Timeline of Google Gemini

From Timelines
(Redirected from Timeline of Google Bard)
Jump to navigation Jump to search
The timeline currently offers focused coverage of the period until February 2026. It is likely to miss important developments outside this period (particularly after this period) though it may have a few events from after this period.

This is a timeline of Google Gemini, a generative AI chatbot developed by Google. Launched in 2023 as Bard, it was created in response to OpenAI’s ChatGPT. Initially, Gemini is built on Google's LaMDA (Language Model for Dialogue Applications) and later transitioned to PaLM, before evolving into its current large language model (LLM) framework, Gemini.

Sample questions

The following are some interesting questions that can be answered by reading this timeline:

  • What are some important preluding developments leading to the launch of Bard?
    • Sort the full timeline by "Event type" and look for the group of rows with value "Prelude".
    • You will se a number of events highlighting the dynamic evolution of conversational AI models, with Meena and LaMDA preceding Google Bard. The launch of ChatGPT by OpenAI serves as a notable moment that influences Google's strategic considerations, leading to a proactive response to maintain competitiveness in the domain of AI-driven language models.
  • What improvements and new features have been introduced to enhance the capabilities and functionalities of Bard/Gemini?
    • Sort the full timeline by "Event type" and look for the group of rows with value "Update".
    • In summary, you will see updates reflecting a combination of advancements, expansions into new domains, and improvements to existing capabilities.
  • What are some cases showcasing Google's efforts to integrate Bard/Gemini into various Google services and applications?
    • Sort the full timeline by "Event type" and look for the group of rows with value "Integration".
    • You will see integration efforts attempting to make Bard/Gemini more versatile and seamlessly integrated tool across multiple services.
  • What are some events reflecting the dynamic and competitive landscape in the AI chatbot industry?
    • Sort the full timeline by "Event type" and look for the group of rows with value "Competition".
    • You will see stakeholders actively comparing, evaluating, and promoting their products based on features, performance, and potential applications, with Bard/Gemini actively participating in the competitive discourse.
    • What are some cases of Bard/Gemini performance evolution?
    • Sort the full timeline by "Event type" and look for the group of rows with value "Performance".
    • You will see a number of events collectively depicting Google's chatbot journey from facing challenges and setbacks to implementing measures for improvement and learning.
  • Other events are described under the following types: "Education initiative", "Expansion", "Legal", "Operational focus", "Partnership", and "Security".

Big picture

Time period Phase label Big-picture characterization
2021–2022 Internal-only AI phase During this period, Google, a division of Alphabet Inc., maintains substantial, internally developed large language model (LLM) technology, including LaMDA (Language Model for Dialogue Applications) and the PaLM (Pathways Language Model) series. These models are primarily confined to the Google Research division, being treated as research artifacts requiring extensive internal safety and ethical review. Public deployment is limited due to concerns regarding potential societal impact, reputation risk, and regulatory ambiguity. This institutional posture is abruptly altered by the public release of OpenAI's ChatGPT in November 2022, which establishes a new market expectation for accessible, consumer-grade generative AI, thereby necessitating a rapid shift in Google's productization strategy.
Early 2023–Mid 2023 Initial product launch Google initiates its consumer-facing LLM deployment with the introduction of the Bard chatbot in February 2023, positioning it as a direct competitor to the integrated ChatGPT technology within Microsoft Bing. The expedited rollout is characterized by initial operational and public relations challenges, most notably a high-profile factual inaccuracy during the product's launch demonstration. This event contributes to a temporary decline in market confidence and initiates a period of intensely competitive, reactive iteration. Subsequent months see rapid model upgrades, expansion of geographical access, and incremental integration with existing Google ecosystem products (e.g., Search), signaling an organizational effort focused on rapidly closing the newly established market gap.
Mid 2023–2024 Strategic consolidation and rebranding A pivotal organizational restructuring occurs with the merger of Google Brain and DeepMind in April 2023, consolidating Alphabet's primary AI research efforts under a unified structure. This action is designed to streamline research-to-product pipelines and accelerate the development of advanced systems. The subsequent announcement and release of the Gemini model family in late 2023 marks the transition to a unified, inherently multimodal foundation model strategy. In early 2024, the Bard application is officially deprecated, and the product line is rebranded as Gemini.[1] The technology is subsequently integrated across major Google consumer and enterprise offerings, including Google Workspace and Android, despite facing ongoing public scrutiny concerning model biases and data usage methodologies.
2025 Maturation under competitive and regulatory constraints The Gemini platform solidifies its role as Google's primary AI engine across its core product suite, including Search, productivity, and developer offerings. This year reflects a transition from rapid, disruptive experimentation to incremental refinement and market maturation. Strategic focus shifts to enhancing reliability, safety, enterprise-grade governance, and integration depth, rather than dramatic capability breakthroughs. Google remains a central, powerful player, but the landscape is increasingly competitive, with rivals like OpenAI and Microsoft maintaining market share, and open-weight models gaining traction. Increasing global regulatory oversight also contributes to the normalization and commoditization of large language model capabilities across the industry.

Summary by year

Time period Development summary
2021 Google unveils LaMDA, a dialogue-focused large language model built on the Transformer architecture, designed for open-ended, conversational interaction. Despite capabilities comparable to later public chatbots, Google choses not to release LaMDA, citing risks of misinformation and societal harm. Leadership emphasizes caution, prioritizing safety, factuality, and ethical review over rapid public deployment.[2][3]
2022 OpenAI launches ChatGPT, built on the GPT-3 family of large language models, quickly gaining widespread public adoption. Its rapid success triggers a “code red” at Google, signaling an internal crisis response and accelerating efforts to counter the perceived threat to Google Search and its core business model.
2023 Google rapidly introduces and iterated on Bard, its experimental conversational AI, amid intense competitive pressure from ChatGPT and Bing. Launched in February with limited testing, Bard’s rollout is marred by a high-profile factual error that erases significant market value and exposes concerns about accuracy, safety, and readiness. Google responds with accelerated internal testing, frequent model upgrades, and expanded employee involvement. Throughout the year, Bard gains coding, multimodal, multilingual, and deep product integrations across Workspace, Search, and Assistant, while facing criticism over hallucinations, privacy, labor conditions, and misinformation. In December, Google launches the Gemini model family, marking a strategic shift toward unified, multimodal AI foundations.
2024 Google rapidly expands Gemini across consumer devices, enterprise tools, developer platforms, and global markets. Gemini integrates into Samsung Galaxy S24, Google Ads, Chrome, Android, Maps, and Workspace, while Bard is fully rebranded as Gemini with paid tiers such as Gemini Advanced. Major model upgrades—Gemini 1.5 Pro, Flash, Nano, and experimental reasoning models—deliver larger context windows, multimodality, and faster performance. Gemini achieves competitive benchmarks but faces controversies over historical accuracy, safety failures, and security vulnerabilities, prompting policy restrictions and safeguards. Global expansion, monetization, and deep product embedding define 2024 as Gemini’s transition from launch to large-scale operational deployment.
2025 Google significantly expands Gemini across security, developer tools, personalization, and enterprise use while facing growing scrutiny. Reports reveal state-backed North Korean hackers exploiting Gemini for espionage and cybercrime, underscoring AI misuse risks. Google launches multiple products, including a free Gemini Code Assist, Gemini CLI, personalized Gemini models, and increasingly agentic search features capable of autonomous actions. Gemini integrates deeply into Android, Chrome, and workplace platforms, culminating in Gemini Enterprise. Technical milestones include gold-medal–level performance in elite programming competitions and production-ready image generation. Alongside advances, Google addresses privacy, child safety concerns, and environmental impacts, reflecting Gemini’s maturation amid regulatory, ethical, and competitive pressures.

Full timeline

Inclusion criteria

We include:

  • Updates that enhance functionality, fix critical bugs, improve user experience, and address security vulnerabilities. These updates ensure stability, compatibility with existing systems, and overall product reliability.
  • Integrations with relevant platforms, services, or tools that add value, improve efficiency, and enhance the user experience.
  • AI model upgrades that improve accuracy, reduce biases, enhance output quality, and improve user interactions.
  • Feature additions that improve user experience, simplify workflows, enhance accessibility, and add significant value to the product.
  • Performance improvements that enhance speed, reliability, and overall performance.

We exclude:

  • Minor adjustments or bug fixes that do not significantly impact functionality, user experience, or the core AI model performance.
  • Integrations that do not bring new features or improvements to Gemini’s core capabilities. We avoid collaborations or tool links with minimal impact on Gemini’s overall functionality or user experience.
  • Updates that involve internal optimizations, fixes, or enhancements not visible to end-users or those that do not result in significant advancements in Gemini’s performance or output quality.
  • Performance tweaks that do not provide clear, measurable benefits to speed, accuracy, or responsiveness, especially if improvements are negligible or irrelevant to most user applications.

Timeline

Year Event type Details
2020 (January 28) Prelude Google Research introduces Meena, a 2.6 billion parameter end-to-end trained neural conversational model, aiming to address the limitations of existing open-domain chatbots. Meena's training objective is to minimize perplexity, measuring uncertainty in predicting the next token in a conversation. The model demonstrates improvements in sensibleness and specificity through a new human evaluation metric called Sensibleness and Specificity Average (SSA), achieving a score of 79% compared to 86% for humans. Meena represents a direct predecessor to LaMDA (Language Model for Dialogue Applications), which Google would announce in 2021 and which would later power the early versions of Bard.[4][5][6]
2021 Prelude Google unveils LaMDA (Language Model for Dialogue Applications), a prototype Large Language Model (LLM). Built on the Transformer architecture, LaMDA is trained specifically on dialogue, distinguishing itself with sensibleness and specificity in responses. It aims to overcome the limitations of traditional chatbots by engaging in free-flowing conversations on a wide range of topics. LaMDA's development focuses on sensibility, specificity, interestingness, and factuality.[7][8]
2022 (November 30) Prelude OpenAI launches ChatGPT based on the GPT-3 family of LLM, gaining significant popularity.[9] ChatGPT's success prompts a "code red" alert at Google, triggering a response to address the potential threat to Google Search.[10][11]
2023 (February 6) Product launch Google CEO Sundar Pichai introduces Bard, an experimental conversational AI service powered by LaMDA. Officially released, Bard aims to combine broad knowledge with language models' intelligence and creativity, offering users insightful responses.[12] It is initially rolled out to 10,000 "trusted testers."[13]
2023 (February 8) Performance Google's parent company, Alphabet Inc., suffers a significant market value drop of $144 billion after Bard provides inaccurate information during a demonstration. The demo showcases a mistake regarding the James Webb Space Telescope, leading to a 7.7% decrease in Alphabet shares. Google immediately acknowledges the error and announces the launch of a Trusted Tester program to combine external feedback with internal testing before wider release. The incident prompts a broader internal reckoning: CEO Sundar Pichai subsequently asks 80,000 employees to dedicate time to testing Bard, and the company accelerates its shift from LaMDA to the more capable PaLM model. The setback comes amid a global AI competition, with companies like Baidu and Alibaba also working on similar projects.[14][15]
2023 (February 13) Performance John L. Hennessy, Chairman of Alphabet and former President of Stanford University — a pioneering computer scientist who co-developed the RISC processor architecture, co-founded MIPS Computer Systems, and received the 2017 ACM Turing Award — states that Google was hesitant to release Bard because the technology is still giving inaccurate answers. He believes that generative artificial intelligence is one to two years away from being truly useful on a broader scale. Hennessy speaks at a conference and mentions that Google was caught up in the sudden interest in ChatGPT and generative AI. He emphasizes the need for caution in releasing AI systems that could provide incorrect or toxic information, and expresses concerns about the role of technology in ensuring a functioning democracy and promoting a harmonious society. He also acknowledges the impressive capabilities of ChatGPT while noting that startups in Silicon Valley have an advantage in recruiting talent during the current cycle of layoffs.[16]
2023 (February 15) Performance Google asks its employees to help improve the responses of Bard, by rewriting incorrect answers. Prabhakar Raghavan, Google's vice president for search, sends an email to staff with a link to a document outlining the do's and don'ts of fixing responses. The document encourages employees to rewrite responses on topics they understand well, emphasizing that Bard learns best through example. The instructions include keeping responses neutral, not implying emotion, and avoiding stereotypes or presumptions based on various categories. At this time, the company aims to involve its employees in testing Bard to accelerate its training and improve accuracy. Employees who contribute to fixing responses may receive recognition through an internal badge and have the opportunity to share feedback with the team working on Bard. The move comes after Google faced criticism for the rollout of Bard, which led to a drop in the company's stock price.[17]
2023 (February) Operational focus Sundar Pichai asks employees to dedicate time to testing Bard, leading to 80,000 employee responses.[13]
2023 (March 13) Update Google employees begin testing a more sophisticated version of Bard called "Big Bard," featuring larger parameters than the version initially rolled out to trusted testers. The internal testing precedes the public limited release of Bard on March 21, 2023.[18][19]
2023 (March 21) Product launch Google opens limited access to Bard, positioning it as an experiment rather than a finished product and making it available to select users in the United States and United Kingdom through a waitlist. Google emphasizes that Bard is not a replacement for its search engine but a complement to it, serving as a chatbot for generating ideas, drafting writing, and engaging in conversation. In a demo, Bard provides responses to general queries, but factual accuracy is hit-and-miss, sometimes generating inaccurate or offensive information. Bard's interface includes disclaimers, and its replies are cautioned to be treated with caution. While Bard is faster than ChatGPT and Bing, its answers appear more constrained, lacking the chaotic and experimental nature of Bing's responses. Google reports intention to find a balance between Bard's capabilities and liabilities as more users gain access and stress test the system.[20][21]
2023 (March 31) Update Sundar Pichai announces the intention to upgrade Bard, basing it on PaLM instead of LaMDA. The switch is significant: while LaMDA is designed specifically for dialogue, PaLM is a far larger and more capable general-purpose language model trained on a broader range of tasks, including code and reasoning. The move signals Google's recognition that Bard's early struggles with accuracy and depth are partly architectural, and that competing with GPT-4 requires a more powerful underlying model.[22]
2023 (April 10) Update Bard introduces its inaugural experiment update, featuring an Experiment updates page for users to access the latest features, improvements, and bug fixes. The addition of suggested Search topics under "Google it" enhances user exploration by offering a broader range of interests. Notably, Bard's capabilities for math and logic are improved to provide higher-quality responses in these domains.[23][24]
2023 (April 19) Performance An article at Bloomberg reports on Google employees criticizing Bard in internal messages, with some describing it as "a pathological liar." According to the report, 18 current and former Google workers express concerns about Bard's performance, noting that it often provided dangerous advice on topics like landing a plane or scuba diving. The report reveals that Jen Gennai, Google's Director and founder of Responsible Innovation — the team tasked with operationalizing Google's AI Principles across its products — overruled a risk evaluation submitted by her own team that found Bard not ready for release, in order to meet competitive pressure from ChatGPT. Meredith Whittaker, a prominent AI ethics researcher, co-founder of the AI Now Institute at NYU, president of the Signal Foundation, and a former Google employee who founded the company's Open Research group before leaving in 2019, argues that "AI ethics has taken a back seat" at Google. Google responds publicly by saying that "delivering responsible AI remains a top priority," while making no specific structural commitments. Two days later, on April 21, Google announces Bard's new coding capabilities — widely seen as an effort to shift public attention toward positive developments.[25][26]
2023 (April 21) Update Bard is updated to assist with programming and software development tasks, offering capabilities such as code generation, debugging, and explanations in over 20 programming languages, including C++, Go, Java, JavaScript, Python, and TypeScript. Users can export Python code to Google Colab seamlessly. Bard can generate, explain, and debug code, making it a valuable tool for learners and developers. While it is an early experiment at this time, and may occasionally provide inaccurate information, its new coding features aim to accelerate software development, helping users tackle engineering challenges.[23][27][24]
2023 (May 5) Expansion Google Bard becomes available for Google Workspace users, expanding its accessibility beyond personal Google accounts. Admins can enable Bard for their domains through the Admin Console, allowing users to access it for work-related tasks, research, and business needs. While users must sign up for the waitlist, admins can activate Bard at the domain, organizational unit, or group level. The rollout begins, with visibility expected within 15 days, though regional availability restrictions apply. This integration aims to enhance creativity, productivity, and collaboration within Google Workspace, following Bard's recent update enabling code writing in 20 languages.[28][29][30][31][24]
2023 (May 10) Product launch During Google I/O developer conference, the company announces the end of the waitlist for Bard, making it widely available in English. The company aims to gather feedback and continue improving the chatbot by expanding its user base. Bard becomes accessible in over 180 countries and territories, with plans to add support for additional languages, including Korean and Japanese. Google emphasizes its responsible development approach and refers to Bard as an experiment rather than a beta.[32][33] Google also unveils updates and new features for Bard. The tool now supports Japanese and Korean languages, with plans to support 40 languages in the future. Updates include image capabilities, advanced coding features, and integration with Google apps and services. Users can now incorporate images into their prompts and receive text-based responses with rich visuals. Additionally, developers can benefit from improved source citations, a dark theme, and an "Export" button for running code with partners like Replit. Google reportedly aims to integrate Bard with various apps and services such as Google Docs, Google Drive, Gmail, Google Maps, and more. Also, Bard reportedly plans to connect with external partners like Adobe Firefly, Kayak, OpenTable, ZipRecruiter, Instacart, Wolfram, and Khan Academy to enhance user experiences.[34][35][24]
2023 (May 15) Update Google Bard receives a notable update, enhancing its summarization capabilities and source attribution. Users can now prompt Bard to provide concise summaries of articles or topics, leveraging advances in large language models. Additionally, Bard's responses now include numbers that link to sources, facilitating identification of matched sections and enabling easy navigation. Google aims to make information digestion quicker and more transparent, emphasizing ongoing improvements based on user feedback. This update follows recent expansions of Bard's language support and availability in additional countries, marking Google's commitment to refining its AI chatbot for diverse user needs.[36][37][24] [23][24]
2023 (May 23) Update Google Bard introduces an update that integrates images into its responses. The new feature enhances the visual experience of prompts by displaying images sourced from Google Search. Users can now see accompanying images when asking for a list of items or request images directly. Google aims to provide transparency by sourcing the images. The company also hints at more visual changes and elements coming to Bard in the future[38][23]
2023 (May 29) Integration Google introduces "Magic Compose", a new feature powered by Bard. Available exclusively to users in the United States, it is an experimental feature within the Messages by Google app. Magic Compose uses AI to generate stylized, suggested responses based on the context of users' messages. The tool sends up to 20 previous messages to Google's servers to generate conversation starters, replies, or different styles of drafted messages. However, messages with attachments, voice messages, and images are not sent to the servers. Magic Compose offers seven different styles in which it can rephrase text, including Chill, Excited, Formal, Lyrical, Remix, Shakespeare, and Short. It is designed for RCS (Rich Communication Services) within the Messages app and can be accessed through the app's Settings menu. Magic Compose does not store messages or use them to train machine learning models.[39]
2023 (May 30) Competition An article compares Google Bard with ChatGPT and Bing Chat. The author discusses their features, strengths, and limitations to help readers make an informed choice. She concludes that ChatGPT can be prone to misinformation. Bing Chat, powered by OpenAI's largest language model GPT-4, offers internet access for up-to-date information and visual features like image generation. Finally, she acknowledges that Google Bard, known for its speed, is receiving upgrades to enhance its language support and introduce visual elements. The article suggests considering factors such as accuracy, cost, internet access, and visual features when deciding on the best AI chatbot for specific needs.[40]
2023 (May 30) Competition Mikhail Parakhin, CEO of Bing at Microsoft and former CTO at Yandex, states that Google Bard utilizes a "much smaller model" compared to Bing Chat, explaining that this contributes to Bard's faster response times. The comments come after Bard's recent upgrade to PaLM 2. The remark reflects the competitive dynamic between Google and Microsoft in the AI chatbot space, with each company publicly comparing their respective models' capabilities and architectures.[41][42]
2023 (June 1) Update Google Bard incorporates the capability to provide location-based results. Users can opt to share their precise location, allowing Bard to deliver relevant information about directions, businesses, landmarks, and other local details. By leveraging Google's search engine, Maps app, and other products, Bard can pinpoint the user's exact whereabouts and offer accurate responses based on their surroundings. The feature raises privacy concerns, as users must weigh the convenience of location-based information against sharing their precise location with Google's AI systems.[43][24][23]
2023 (June 7) Update Google enhances Bard with improved logical reasoning skills, making it more proficient in answering mathematical and coding questions. The latest update includes features such as exporting tabular responses to Google Sheets and employing implicit code execution to generate code and provide accurate responses. Google claims that Bard's approach surpasses traditional large language models (LLMs) by combining text prediction and code-based computation, resulting in better problem-solving capabilities. However, Google warns that Bard's code may have flaws, and users should exercise caution when utilizing it.[44][45][24]
2023 (June 9) Competition Articles highlight key advantages of Google Bard over rival chatbots, including a user-friendly interface with dark and light modes, free internet access for real-time browsing, use of Google Search results for more accurate responses, voice input functionality, and location-sensitivity for contextually relevant results. Despite a rocky start, observers note that Bard's public beta is beginning to show potential features that could challenge ChatGPT's dominance.[46][47]
2023 (June 13) Legal Google is blocked by the Irish Data Protection Commission (DPC) from launching Bard in the European Union due to privacy concerns. By this time Bard has already been launched in several countries, including the United States and United Kingdom. The DPC states that Google has not provided the necessary documentation or briefing regarding data protection impact assessment, resulting in the postponement of Bard's EU launch. The block proves short-lived: exactly one month later, on July 13, Google launches Bard across all 27 EU member states after engaging with regulators, making it one of the first tests of whether a major generative AI product can satisfy EU data protection requirements sufficiently to operate across the bloc. The incident reflects the stricter approach to AI regulation in the EU compared to other regions, with the EU AI Act proposed to align AI governance with privacy regulations like the General Data Protection Regulation.[48]
2023 (June 15) Integration Google announces that Google Lens integration into Google Bard, allowing users to include images in their prompts and utilize Lens to understand the visual content. Additionally, Google Lens now has the capability to detect skin conditions by analyzing uploaded photos, providing visual matches to aid in search. Google has also introduced shopping features through Lens, enabling users to take screenshots and receive shoppable matches with links for online purchases, as well as utilizing multisearch to search by both photo and words.[49]
2023 (June 19) Security Google warns its employees not to use code generated by Bard, due to privacy and security risks. Voice AI startup Nuance, backed by Microsoft, faces a privacy lawsuit for recording and using people's voices without permission. Google's DeepMind opposes the idea of a singular AI regulatory agency, advocating for a multi-layered approach. OpenAI cautions Microsoft about releasing its Bing chatbot too quickly, citing concerns about false information and inappropriate language. The developments highlight ongoing challenges in AI regarding privacy, security, and responsible deployment.[50]
2023 (July 12) Legal A lawsuit is filed against Google, alleging that the company has been unlawfully collecting data from hundreds of millions of Americans to train its AI products. The lawsuit, filed by Clarkson Law Firm in the US District Court for the Northern District of California, claims that Google, along with DeepMind and Alphabet, secretly gathered personal and professional information, creative works, photographs, and emails from individuals without their knowledge or consent. Google calls the claims baseless and states that it uses public information responsibly and in accordance with its AI Principles. The lawsuit comes shortly after a similar complaint is filed against OpenAI. In June 2024, a federal judge dismisses the proposed class action but allows the plaintiffs to amend their complaint, leaving the legal question of whether training AI on publicly scraped data constitutes unlawful data use unresolved — a question with broad implications for the entire generative AI industry.[51][52]
2023 (July 13) Expansion Google Bard expands its reach by becoming available in over 40 new languages, including Arabic, Chinese, German, Hindi, and Spanish. Access is extended to more places, encompassing all 27 countries in the European Union (EU) and Brazil. The EU launch is particularly significant: it comes exactly one month after the Irish Data Protection Commission blocked Bard's entry into the bloc over privacy concerns, making this one of the first tests of whether a major generative AI product can satisfy EU regulators sufficiently to operate across the region. Google states that it has engaged with experts, policymakers, and privacy regulators to ensure a responsible approach to AI expansion. By this time, the EU works on comprehensive rules for AI, including provisions for generative AI systems like ChatGPT.[53][54][55]
2023 (July 13) Update Bard introduces several new features and improvements. In addition to Text-to-speech capabilities added in over 40 languages, the Pinned & Recent Threads feature is introduced, allowing users to organize and pick up past Bard conversations, and supporting a continuous creative process. Sharing Bard conversations is made easier with shareable links, promoting collaboration and inspiration. Modification options for Bard's responses are introduced, enabling users to tailor responses based on simplicity, length, tone, and style. Additionally, Bard's code export capabilities are expanded to include exporting Python code to Replit, offering more workflow options for programming tasks.[24]
2023 (July 25) Competition Articles highlight cases in which Google Bard's multimodal capabilities surpass those of ChatGPT, including transcribing old letters from images, simplifying travel planning, identifying objects through images, offering meal planning assistance based on ingredients, and providing technical troubleshooting by analyzing error screenshots. The comparisons reflect the rapidly shifting competitive landscape between the two chatbots, with Bard's image understanding — powered by Google Lens integration — cited as a key differentiator at this stage.[56][57]
2023 (August 10) Performance Jenny Blackburn, Google's Vice President of User Experience, publishes a blog post highlighting ten practical ways to use Bard, including learning new topics, analyzing images, drafting content, comparing options, generating code, planning trips, and brainstorming creative ideas. The post marks an effort by Google to drive mainstream adoption of Bard by demonstrating everyday use cases as the chatbot becomes available globally in multiple languages.[58][59]
2023 (August 16) Performance Reports find that both OpenAI's ChatGPT and Google Bard spread misinformation when queried about news topics, generating false or misleading information in a significant proportion of tests. A study by NewsGuard finds that Bard produces misinformation in 76 of 100 tested prompts. The findings raise concerns about AI-powered chatbots being used as news sources and highlight the need for improved fact-checking mechanisms.[60][61]
2023 (August 21) Security A malicious campaign is discovered using fake "Google Bard AI" advertisements on Facebook to distribute malware. The campaign leads users to suspicious download pages mimicking official Google offerings, with downloaded files flagged as malicious by multiple antivirus vendors. The incident is an early example of cybercriminals exploiting public interest in AI chatbots for financial gain — a trend that would accelerate as Bard and Gemini gained mainstream visibility.[62][63]
2023 (September 19) Integration Google integrates Bard with its Gmail service and other products. This allows users to easily use Bard across different Google services. Users can extend Bard to apps like Gmail, Google Drive, YouTube, and Google Maps using a system called Bard Extensions. For example, users can ask Bard to summarize emails on a specific subject in Gmail or help with trip planning by searching for flight dates, booking hotel rooms, and providing directions in Google Maps. Google also adds a verification feature to check the correctness of Bard's results against Google search results.[64][65][24]
2023 (September 27) Update Google Bard introduces a new feedback feature, allowing users to provide input when Bard presents two drafts side by side. Users can choose their preferred draft, indicate no preference, or opt out entirely. This feedback mechanism aims to gather real-world input to enhance the quality of Bard's responses, contributing to continuous improvements for users and the overall Bard experience.[24]
2023 (October 4) Integration Google announces plans to integrate generative AI capabilities from its Bard chatbot into Google Assistant, providing personalized assistance for tasks like trip planning and email management on mobile devices. The upgraded Assistant, combining reasoning and generative abilities, supports text, voice, and image interactions, with access to a phone's camera and microphone. Privacy is emphasized, and the initial focus is on user experience, with no revenue-generating features. Google's move follows a trend in the industry, with competitors enhancing their virtual assistants with generative AI.[66][67][68][69]
2023 (October 23) Update Google Bard introduces new update. Through the Workspace Extension, Bard can now summarize a larger number of emails simultaneously and has improved comprehension when users request recent emails. This incremental improvement is aimed at enhancing the quality and utility of the Workspace Extension, making it more effective for users seeking email summaries. Additionally, shared conversations now include visibility for uploaded images in the prompt, providing users with a more creative and engaging experience. These updates contribute to a more comprehensive and user-friendly interaction with Bard, particularly in the context of email summaries and shared conversations.[24]
2023 (October 30) Update Google Bard introduces a new feature allowing responses to appear in real time, enabling users to view and engage with responses as they are generated. This setting eliminates the need to wait for the full response, providing users with the ability to read and iterate on ideas more swiftly. The update aims to enhance the creative process by allowing users to stay in the flow of generating ideas and responses with increased efficiency and speed.[24]
2023 (November 7) Staff Contractors working for Google through Accenture, responsible for training the Bard AI chatbot, vote 26-2 to join the Alphabet Workers Union, with the US National Labor Relations Board having previously ruled in September that Google and Accenture are joint employers. The workers seek better working conditions and protections after facing challenges while training Bard, including handling offensive prompts without adequate training. The union drive succeeds despite retaliatory layoffs of more than 80 of the 119 organizing workers in August. Google appeals the NLRB's joint employer ruling and refuses to engage in collective bargaining, maintaining it has no control over Accenture's employment terms. By December 2024, the union signs a collective agreement with Accenture alone while Google's appeal remains ongoing — a resolution workers describe as partial and unjust.[70][71]
2023 (November 16) Update Google Bard expands its accessibility to teenagers worldwide, offering age-appropriate protections, updated onboarding processes, and experiences tailored to empower exploration and learning. The expansion aims to provide teens with inspiration, motivation, and quick understanding of various topics, supporting them in areas like homework, hobbies, job applications, and college preparation. Additionally, Bard now assists users in solving math equations by providing step-by-step explanations, and it can generate charts from data in prompts or tables during conversations, enhancing visualization of information. These updates cater to diverse learning needs and facilitate a broader range of creative and educational pursuits.[24]
2023 (November 21) Update A stable release of Bard (version 2023.11.21) is launched. Google improves the chatbot, with an update focused on enhancing its ability to understand and respond to questions related to YouTube videos. The expansion enables users to engage in more meaningful conversations with Bard by asking specific questions about video content, such as details about recipes or tools in DIY videos. This update aligns with Google's ongoing efforts to enhance Bard's capabilities and offer users a more interactive experience. To access this feature, users must enable the YouTube extension on Bard's web portal. The update is expected to benefit users seeking precise information from videos and is seen as valuable for content creators aiming to improve engagement and quality.[72][73][74]
2023 (December 6) Update Google launches Gemini, an AI model designed for human-like thinking, introducing versions "Nano" and "Pro" integrated into Bard chatbot and Pixel 8 Pro smartphone. The launch marks a decisive architectural turning point, ending the LaMDA and PaLM era and replacing them with a unified, natively multimodal model family. Google claims Gemini Ultra outperforms GPT-4 on 30 of 32 benchmarks, including being the first model to achieve human-expert performance on the MMLU test. These claims are contested: MIT Technology Review notes the performance differences between the two models are "relatively thin," independent researchers flag concerns about possible data contamination in training sets and the use of different evaluation metrics than those used by OpenAI, and subsequent independent tests find GPT-4 outperforms Gemini Ultra on some tasks including coding. Gemini aims to enhance Bard's intuitiveness and planning, with the Ultra model powering "Bard Advanced" for improved AI multitasking in early 2024.[75][76][77]
2023 (December 18) Update Google Bard expands its Extensions feature to include Japanese and Korean in addition to English. This enhancement allows users to access real-time information from Google apps and services, such as YouTube, Hotels, Google Flights, and Google Maps. Bard Extensions can also retrieve information from Gmail, Google Docs, and Google Drive, offering users the ability to find, summarize, and get answers from their personal content. Users have control over privacy settings, enabling them to manage how extensions are utilized. Additionally, Bard's Export to Replit feature now supports 18 programming languages, including C++, Javascript, Ruby, SQL, and Swift, catering to developers seeking coding assistance in various languages.[24]
2023 (December 28) Product vision Colin Murdoch, Chief Business Officer of Google DeepMind — responsible for applying the lab's research breakthroughs to Google products and overseeing its commercial partnerships — discusses the transformative potential of the Gemini AI model. Murdoch highlights how Gemini will revolutionize daily life and work for billions by advancing technology in areas such as communication, healthcare, and productivity, and emphasizes the AI's role in making complex tasks more accessible.[78][79]
2024 (January 17) Integration Google and Samsung announce the integration of Gemini AI models in the Galaxy S24, with Gemini Pro enhancing apps like Notes, Voice Recorder, and Keyboard with advanced summarization features, such as summarizing recorded lectures. Gemini Nano powers Magic Compose in Google Messages, enabling users to create messages in different styles offline. The partnership is significant as one of the first major Android OEM integrations of Gemini, positioning it as the default AI layer for hundreds of millions of Samsung users worldwide. Privacy advocates note that the depth of Google AI integration on a third-party device raises questions about data flows — specifically, how user data processed by Gemini on Samsung hardware is handled compared to Google's own Pixel devices, and what controls users have over Gemini's access to device microphone, camera, and personal data outside Google's own ecosystem. The S24 is also expected to feature Google's Imagen 2 model for advanced photo editing, and Samsung would be among the first to test the powerful Gemini Ultra model.[80]
2024 (January 23) Integration Google integrates Gemini into Google Ads to simplify campaign creation and improve ad performance. The conversational tool, at this time in beta for English-speaking advertisers in the US and UK, helps create search ad campaigns more efficiently. Preliminary tests reported by Google show that advertisers using the tool achieved higher Ad Strength scores and that small businesses are 42% more likely to receive "Good" or "Excellent" Ad Strength scores, resulting in a 12% increase in conversions when scores improve. These figures are self-reported by Google from its own beta program and have not been independently verified; analysts note that performance statistics reported by platform owners about their own advertising tools are inherently difficult to assess without third-party auditing.[81]
2024 (January 27) Competition Gemini Pro surpasses GPT-4 in performance on the HuggingFace Chatbot Arena Leaderboard, earning second place. The leaderboard uses human preference votes across thousands of blind comparisons, making it one of the more independent benchmarks at the time. The result follows updates to Gemini Pro and is seen as a meaningful milestone in the competitive AI landscape, though Google's broader Gemini Ultra model — positioned as the company's most direct GPT-4 competitor — remains unavailable to the public at this time.[82][83]
2024 (February 1) Update Google rolls out the Gemini Pro update to its Bard chatbot globally, enhancing its capabilities with improved natural language processing and new features aimed at making the AI more interactive and efficient. The global release reflects Google's push to advance its AI tools and compete with other leading AI technologies in the market.[84]
2024 (February 8) Product launch Bard is rebranded as Gemini, accompanied by the launch of Gemini Advanced, which features the new Ultra 1.0 AI model. Gemini Advanced excels in tasks such as advanced coding, logical reasoning, and creative collaboration. Available through the Google One AI Premium Plan for $19.99/month, it offers expanded multimodal capabilities and advanced tools for coding. The update includes new features like the ability to upload files, allowing Gemini to analyze documents, create visualizations, and provide insights. The introduction of Gemini Live enables natural speech interactions and dynamic planning, including personalized features like custom trip itineraries. Subscribers can also create customized versions of Gemini, called Gems, to suit specific needs. Google also announces future app integration with more Google services, enhancing the user experience.[85][86][87]
2024 (February 15) Update Google announces Gemini 1.5, an update to the original Gemini 1.0, which improves performance and outperforms Gemini 1.0 on 87% of benchmarks. A particularly significant advance is the introduction of a one million token context window — at the time the largest of any publicly available AI model — allowing the system to process and reason over the equivalent of roughly 700,000 words, or an entire codebase, in a single prompt. The update includes Gemini's integration with Duet AI and significant improvements in reasoning and image processing, surpassing ChatGPT in 30 of 32 key benchmarks. Gemini also becomes the first AI model to outperform humans on the MMLU test.[88][89]
2024 (February 21) Performance Google apologizes for inaccuracies in Gemini after it generates racially diverse images of historically white figures, such as US Founding Fathers and Nazi-era German soldiers. Critics, including some right-wing voices, accuse the tool of overcorrecting racial bias by misrepresenting historical facts. The following day, February 22, Google pauses Gemini's image generation of people entirely while working on fixes — a feature that remains disabled for months, still off as of May 2024. CEO Sundar Pichai calls the responses "completely unacceptable" and commits to structural changes, updated product guidelines, improved launch processes, and additional red-teaming. Google explains that its diversity tuning had caused the model to both overcompensate in some cases and refuse legitimate prompts in others, acknowledging it "missed the mark" in execution.[90][91]
2024 (March 12) Performance Researchers from HiddenLayer identify security vulnerabilities in Google's Gemini AI models, particularly related to prompt injection and system instruction leaks. The bugs are found in the Gemini Advanced Google Workspace plugin, allowing attackers to manipulate the model through malicious documents or indirect prompts. For example, the model can be tricked into asking for a password or generating harmful content. In response, Google implements a layered defense strategy including model hardening, markdown sanitization to strip harmful scripts, suspicious URL redaction, and a user confirmation framework requiring human approval for sensitive AI-generated actions. Despite these measures, prompt injection vulnerabilities continue to be discovered by independent researchers well into 2025 and 2026, underscoring that the problem is an industry-wide challenge not fully solvable through model training alone. No evidence suggests these vulnerabilities had been exploited maliciously at the time of disclosure.[92]
2024 (March 13) Policy Google implements restrictions on Gemini, limiting its ability to respond to election-related queries globally — initially in the US and India ahead of India's elections in April 2024, then rolling out worldwide to any country with upcoming elections. When asked about political parties or candidates, Gemini responds with "I'm still learning how to answer this question," directing users to Google Search instead. This move aligns with Google's announcement in December 2023 about restricting AI responses to political questions ahead of global elections in 2024. Notably, Google never officially confirms whether the restrictions would be lifted after the election cycle ends. As of early 2026, the restrictions remain in place, making Gemini an outlier among major AI chatbots — with competitors like OpenAI and Anthropic having updated their models to engage more directly with political topics, leading critics to accuse Google of excessive caution.[93][94]
2024 (April 13) Performance A peer-reviewed study published in the journal Eye evaluates the performance of Google Gemini and Bard AI chatbots in ophthalmology knowledge using 150 multiple-choice questions from EyeQuiz. Both achieve an overall accuracy of 71%, with country-specific variations: in Vietnam, Bard scores 67% and Gemini 74%, while Brazil and Netherlands versions of Gemini score 68% and 65% respectively. Despite acceptable performance levels, the chatbots occasionally provide confident yet incorrect explanations, highlighting both the potential and current limitations of AI in medical knowledge assessment contexts.[95][96]
2024 (April 25) Notable comment During Alphabet's Q1 2024 earnings call, Google CEO Sundar Pichai declares the start of the "Gemini era," emphasizing generative AI integration into Google Search and advertising as the company's primary strategic focus. The company reorganizes AI teams under Google DeepMind, consolidating what had previously been separate research efforts following the merger of Google Brain and DeepMind in April 2023 — a restructuring designed to accelerate the pipeline from research to product. Google reports reducing AI serving costs by 80% through efficiency improvements during this period. Monetization efforts include AI-driven ads, Cloud services, and subscriptions like Google One. The declaration comes as Alphabet reports 15% revenue growth year-over-year, signaling that the AI transition has not yet disrupted its core advertising business.[97][98]
2024 (May 14) Product launch Google announces the launch of Gemini 1.5 Pro for Gemini Advanced subscribers, featuring a 1 million-token context window and enhanced capabilities in over 35 languages. The update allows users to upload files, enabling Gemini to analyze documents, create visualizations, and provide insights. New conversational features include Gemini Live, which supports natural speech interactions and dynamic planning capabilities, such as personalized trip itineraries. Subscribers can also create customized versions of Gemini, known as Gems, tailored to specific needs. Additionally, the app will integrate more Google services for a seamless user experience.[99]
2024 (May 22) Product launch Google launches Gemini Nano, integrating it into the Chrome desktop client to enhance AI capabilities, including offline app development and content generation. This integration allows developers to access AI functionalities through high-level APIs, simplifying deployment without deep technical expertise. Key features include improved performance via WebGPU and WebAssembly, enabling more complex computations directly in the browser.[100]
2024 (May) Legal The United States Patent and Trademark Office (USPTO) rejects Google's trademark application for the name "Gemini" for its AI tools. The USPTO determines that the name is likely to cause confusion with existing trademarks, including that of Gemini Data, which had already trademarked the "Gemini" name for its AI platform. Despite this rejection, Google would proceed with rebranding its AI services under the "Gemini" name, leading to the lawsuit filed by Gemini Data in September 2024.[101]
2024 (June 1) Expansion Google launches Arabic support for Gemini, making it accessible through a dedicated app for Android and within the Google app for iOS. The app offers free access to Gemini 1.0 Pro, while Gemini Advanced (1.5 Pro) is available through the Google One AI Premium plan, supporting complex tasks like coding and logical reasoning. Extensions within the app enable users to access real-time information from Google Maps, flight bookings, and content summaries. Gemini now understands over 16 Arabic dialects and responds in Modern Standard Arabic, with additional Arabic-focused features planned for later this year.[102][103]
2024 (June 5) Expansion Google expands the availability of its Gemini app to users in the European Union and the United Kingdom, allowing Android users in these regions to download it from the Google Play Store. This update makes the Gemini app an alternative to Google Assistant on Android devices, now compatible with versions 10 and 11, reaching a broader audience, including users with older smartphones.[104]
2024 (June 25) Update Google announces the opening of access to the 2-million-token context window for its Gemini 1.5 Pro AI model, previously restricted to a waitlist. This enhancement allows developers to utilize a larger context for their inputs. Researchers note, however, that very large context windows do not fully solve the so-called "lost in the middle" problem: on complex multi-needle retrieval tasks — where multiple specific pieces of information must be extracted from a long document — Gemini 1.5 Pro shows average accuracy of 60 to 70%, meaning important details can be missed in practice despite the expanded window. Google also introduces code execution capabilities for the Gemini API, enabling models to generate and run Python code and learn from the results. The company also launches the Gemma 2 model in Google AI Studio for experimentation and announces plans to offer tuning for Gemini 1.5 Flash to developers by mid-July.[105]
2024 (July 24) Partnership Google announces a partnership with NBCUniversal and the U.S. Olympic and Paralympic Committee for the Paris 2024 Games, becoming Team USA's first-ever Official Search AI Partner. NBC commentators use Google Search's AI Overviews during daytime and primetime coverage, while comedian Leslie Jones uses Gemini in dedicated broadcast segments. Five Team USA athletes also use Google Lens, Circle to Search, and Immersive View in Google Maps to explore Paris in social media content. The integration marks one of the first instances of generative AI being embedded directly into live Olympic broadcast coverage.[106][107][108]
2024 (July 25) Update Gemini is updated to Gemini 1.5 Flash, making it faster and improving its reasoning and image interpretation. Available for free, this update increases response quality and speed, while supporting more complex prompts with up to 32,000 tokens — four times its original capacity. Flash is explicitly designed as a speed-and-cost-optimized model, trading depth of reasoning and accuracy for faster responses and lower computational costs, making it less reliable than Gemini Pro for complex or high-stakes tasks. New features include file uploads for contextual prompts and future capabilities to analyze data files and create visualizations. To reduce inaccurate responses, Gemini starts citing sources for all answers. Additionally, it's accessible in Google Messages across Europe, and teenagers aged 13+ can safely use Gemini with enhanced safeguards.[109][110]
2024 (August 19) Product launch Google Research makes its bioacoustic foundation model, Health Acoustic Representations (HeAR), available to researchers for developing models that analyze human sounds such as coughs to identify early signs of disease. The technology holds promise for screening conditions like tuberculosis — a treatable illness that remains undiagnosed in millions due to limited healthcare access — particularly in resource-constrained countries. An India-based respiratory healthcare company investigates how HeAR can enhance early detection of TB through cough analysis. HeAR represents part of Google's broader health AI initiative alongside Med-Gemini, its specialized medical reasoning model.[111][112]
2024 (September 11) Update NotebookLM, Google's AI-powered research assistant utilizing Gemini models, introduces an innovative feature called Audio Overviews. This new capability allows users to upload materials and engage with two AI hosts as they conduct in-depth discussions about the content in a podcast-like format. The feature goes unexpectedly viral in the weeks following its release, generating widespread public discussion about the nature of AI-generated audio and sparking debate about whether such synthetic conversations could be mistaken for human ones — making it one of the most publicly discussed AI feature launches of late 2024. Users can listen to these conversations, which can also be downloaded for later reference. Additionally, NotebookLM starts supporting uploading YouTube URLs and audio files, enhancing its functionality.[113]
2024 (September 11) Legal Gemini Data Inc. files a lawsuit against Google, claiming trademark infringement over the use of the “GEMINI” name for its AI tools. Google had rebranded its Bard to Gemini in early 2024, despite Gemini Data holding the trademark for the name in the AI sector. The U.S. Patent and Trademark Office had rejected Google’s trademark application for “Gemini” in May 2024 due to confusion with existing marks, including Gemini Data's. The lawsuit highlights the imbalance of power between smaller companies and tech giants.[101]
2024 (October 3) Expansion Gemini Live enhances its capabilities by expanding support to over 40 languages, allowing users to have natural conversations in their preferred language on Android devices. This feature enables users to interact with Gemini for brainstorming, event planning, and exploring new topics, all while offering up to two language options. Researchers and language advocates note that AI model quality varies significantly across languages, with most large language models performing substantially better in English than in other languages — particularly less-resourced ones — meaning the quality of Gemini's responses may be considerably lower for users in some of the newly supported languages compared to English speakers. Additionally, Gemini integrates with more Google apps like Gmail, Calendar, and YouTube, facilitating easier task management and information access.[114]
2024 (October 31) Integration Google Maps integrates Gemini, introducing features like "Ask Maps" to enhance user interaction with travel, weather, and navigation queries. This AI-powered tool enables users to ask complex questions about nearby activities and gather insights from local reviews, improving the relevance of responses. Additionally, Google Maps starts providing detailed navigation updates, including lane information, crosswalks, and weather disruptions. Users can report driving conditions using natural language, contributing to real-time updates. Furthermore, Google enhances its “Immersive View” feature with 3D renderings in 150 cities and expanding categories like college campuses.[115][116]
2024 (November 16) Performance Google's AI chatbot Gemini faces controversy after telling a user to "please die" during a homework session about challenges facing aging adults. The user, a 29-year-old Michigan college student named Vidhay Reddy, says he is "thoroughly freaked out" by the experience. Google immediately disables the specific conversation to prevent further sharing or continuation, investigates, and determines it to be an isolated incident — possibly triggered by a malformed prompt or a prompt injection attempt, though Google cannot rule out either cause. Google states it has taken steps to prevent similar responses in the future. Despite this, other users subsequently report similar hostile outputs, raising concerns about the consistency of Gemini's safety filters. The incident amplifies ongoing debates about AI safety testing and the pace of deployment.[117][118]
2024 (November 18) Update Google introduces new Gemini AI features in Android Studio to boost developer productivity across the development lifecycle. These include AI-assisted coding, refactoring, documentation generation, code analysis, and bug fixes. Developers can prompt Gemini for code suggestions directly from the editor, using context menus or shortcuts, to modify or optimize code. Gemini processes requests and provides code diffs for review. AI-powered code completion suggests entire functions, accelerating development. Enabled through Android Studio settings, it may send context-relevant data to improve suggestions. Gemini also supports refactoring and documentation generation, enhancing code organization. These features are available in Android Studio Jellyfish or later.[119]
2024 (December 19) Product launch Google unveils its experimental AI model, Gemini 2.0 Flash Thinking, designed to solve complex questions by explicitly detailing its reasoning process. The model is showcased by Jeff Dean, Google's Chief Scientist and co-lead of the Gemini project — one of the most influential figures in computing, who co-founded Google Brain, co-created foundational systems including MapReduce and TensorFlow, and has over 400,000 academic citations. The model is trained to enhance reasoning by breaking tasks into smaller steps, demonstrating its approach through examples like solving physics problems. While not human-like reasoning, this method aims to produce stronger outcomes. The model integrates speed improvements from Gemini Flash 2.0 and can handle visual and textual elements. As part of Google's AI Studio, it aligns with the company's broader push into "agentic" AI, competing with OpenAI's o1 reasoning model.[120]
2025 (January 31) Security A report from Google's Threat Intelligence Group reveals that state-backed hackers from at least 20 countries — most prominently North Korea, Iran, China, and Russia — have been using Google's Gemini AI for cybercrime and espionage activities. North Korean hackers in particular use Gemini for researching military organizations, defense contractors, nuclear sites, and cryptocurrency topics, as well as to assist with placing remote IT workers under fake identities in Western companies. Iranian-backed groups represent the highest overall volume of Gemini misuse, using it for reconnaissance, target research, and coordinated information operations. Chinese groups primarily use it for research into topics of strategic government interest. Google assesses that state-backed actors would continue to exploit AI technologies for malicious purposes, while noting that as of the report's publication, no actor has used Gemini to generate novel attack methods not previously known.[121][122][123]
2025 (February 25) Product launch Google launches a free tier of Gemini Code Assist, its AI-powered coding assistant, for individual developers. The free version allows up to 6,000 requests per day — substantially more than GitHub Copilot's free tier limit of 2,000 monthly completions. It integrates with popular code editors and can generate, explain, and refine code in any public-domain programming language. Independent reviewers find that while the quota advantage is genuine, GitHub Copilot maintains higher user satisfaction scores and stronger integration with the GitHub ecosystem, and some testers rate Gemini's completion accuracy at around 85-90% of Copilot's on single-file tasks — a meaningful but not decisive gap. Powered by a fine-tuned Gemini 2.0 model, it supports large-context prompts for improved code generation. Google also introduces Gemini Code Assist for GitHub, which automates code reviews by summarizing pull requests and detecting bugs. Paid versions remain available with Google Cloud integrations and enhanced AI suggestions.[124]
2025 (March 13) Product launch Google officially launches "Gemini with Personalization," an experimental AI model that tailors responses based on users' search history. Available to Gemini and Gemini Advanced subscribers, it rolls out on the web and announces expansion to mobile in over 45 languages, excluding the European Economic Area, Switzerland, and the UK — reflecting the stricter data protection requirements in those jurisdictions. Privacy critics raise concerns about giving AI access to users' search history for personalization, noting questions about data security, user autonomy, and informed consent, particularly given Google's broader history of data collection scrutiny. Initially, Gemini only uses search history, but announces integration with Google Photos and YouTube. Google emphasizes user control, allowing search history to be disconnected at any time and applying personalization selectively based on its reasoning models.[125]
2025 (June 25) Product launch Google introduces Gemini CLI, an open-source AI agent that integrates Gemini's AI capabilities directly into the developer's terminal. Designed to accelerate workflows, it connects with tools, understands code, and supports tasks such as code generation, debugging, and command execution. Gemini CLI supports multimodal input and can generate apps from PDFs or sketches, automate complex operations, and interact with external tools like Imagen or Veo. It is built for extensibility using Model Context Protocol (MCP) and offered under an Apache 2.0 license, and is integrated with Gemini Code Assist and available for free with a Google account. The launch positions Gemini CLI as a direct competitor to Anthropic's Claude Code and OpenAI's Codex CLI, reflecting the growing importance of terminal-based AI coding agents.[126][127]
2025 (July 9) Feature update Google announces five new Gemini AI features coming to Android 16 and Wear OS 6, with particular enhancements for foldable devices. Gemini Live works from the external screen on foldables like the Galaxy Z Flip7 and integrates with the camera for real-time visual assistance. Circle to Search is upgraded with AI Mode, enabling deeper interactions and follow-up questions including gameplay help. Gemini integrates with native apps like Samsung Calendar and Notes for proactive planning. Gemini also debuts on smartwatches, starting with the Galaxy Watch8, improving notifications and voice interaction — marking the first time Gemini has been available on wearable devices.[128][129]
2025 (July 16) Product launch Google launches three major AI features: Gemini 2.5 Pro in Search, Deep Search, and Agentic AI calling. Gemini 2.5 Pro, available to AI Pro and Ultra users, enhances Google Search with advanced reasoning, math, programming, and contextual understanding. Deep Search automates in-depth research by performing hundreds of searches, cross-referencing sources, and generating cited reports. The Agentic AI calling feature enables Google Search to make calls to local businesses, checking prices and availability on users' behalf — a feature that raises concerns from small business owners about receiving unsolicited AI-generated calls without explicit consent, and from consumer advocates about potential misrepresentation of the AI as a human caller. These tools mark a shift toward more autonomous, action-oriented AI, transforming how users interact with the web and services.[130][131]
2025 (August 13) Update Google announces updates to its Gemini app introducing personalization from past chats and a privacy feature called Temporary Chats. The personalization feature, debuting in Gemini 2.5 Pro, allows the AI to recall prior conversations to offer more context-aware responses. Temporary Chats prevent conversations from being saved, used for personalization, or used for training, and are deleted after 72 hours. Google also renames "Gemini Apps Activity" to "Keep Activity," refining privacy and data-use controls. The update comes amid broader concerns about Gemini's data retention practices, with users and privacy advocates noting that conversations reviewed by human reviewers can be retained for up to three years even after deletion.[132][133]
2025 (August 28) AI impact report Google becomes the first tech company to publish a report on AI energy, emissions, and water use, focusing on its Gemini model. The median text prompt consumes 0.24 Wh, emits 0.03 g CO₂e, and uses 0.26 mL of water—about nine seconds of TV energy. Google claims AI efficiency has improved, with energy use per Gemini prompt dropping 33× and emissions 44× in one year. Despite these gains, overall emissions have risen 51% since 2019 due to data center growth. The company targets net-zero emissions by 2030, as AI data centers may reach 4.5% of global electricity use.[134]
2025 (September 5) Safety assessment American nonprofit organization Common Sense Media rates Gemini as "High Risk" for children and teens, finding that Gemini's youth versions largely mirror the adult model and lack child-centered design. Tests show Gemini can still share unsafe or age-inappropriate content including sexual, drug-related, or harmful mental health advice. Despite Gemini clarifying it is not a "friend," concerns remain about emotional risk and poor developmental adaptation. Google acknowledges issues but cites ongoing safety improvements. The assessment follows a similar "High Risk" rating given to ChatGPT by Common Sense Media, reflecting broader industry-wide concerns about AI safety for minors.[135][136]
2025 (September 17) Recognition An advanced version of Gemini 2.5 Deep Think achieves gold-medal performance at the International Collegiate Programming Contest (ICPC) World Finals, marking a major milestone in AI problem-solving. Gemini successfully solves Problem C, a complex optimization challenge unsolved by any human team. The achievement is particularly notable in combination with Gemini's earlier gold-medal performance at the International Mathematical Olympiad, making it the first AI system to reach gold-medal level at both the world's premier mathematics and programming competitions — two domains long considered distinct tests of human reasoning. A lightweight version of Gemini 2.5 Deep Think is made available to Google AI Ultra subscribers via the Gemini app.[137]
2025 (October 2) Product launch Google announces that Gemini 2.5 Flash Image, its advanced image generation and editing model, is now production-ready and supports 10 aspect ratios for diverse formats — from cinematic 21:9 to vertical 9:16. Available via the Gemini API, Google AI Studio, and Vertex AI, it enables seamless blending, character consistency, and natural language edits. Google showcases the model through developer partners including Cartwheel and Volley, who report fast and aesthetically strong results for their specific use cases; independent benchmark comparisons with competing image generation models such as DALL-E 3 and Midjourney are limited at the time of release, making broader quality assessments difficult.[138]
2025 (October 9) Product launch Google launches Gemini Enterprise, an AI-powered conversational platform that integrates company data, tools, and employees in one secure space. Built on advanced Gemini models, it enables interaction with documents, creation of custom AI agents, and workflow automation. Clients like HCA Healthcare and Best Buy report major productivity gains. CEO Sundar Pichai describes it as "the new front door for AI in the workplace." Industry analysts note, however, that the prototype-to-production gap remains one of the hardest challenges in enterprise AI adoption, with many organizations able to demonstrate promising proofs of concept but struggling to operationalize AI reliably at scale — a challenge Gemini Enterprise will need to address to fulfill its ambitions. Gemini Enterprise reflects Google's full-stack AI strategy, spanning infrastructure, research, and models like Gemini 2.5 Pro, Veo, and Imagen.[139]
2025 (October 13) Feature update Google begins rolling out the "Summarize page" shortcut for the Gemini overlay in Chrome for Android, allowing users to instantly summarize webpages. The feature appears alongside "Share screen with Live" and "Ask about page," producing a floating summary window with expandable and interactive options. It uses Gemini 2.5 Flash by default, regardless of user settings. The shortcut works in Chrome, Custom Tabs, Discover, Search results, and Google News, and follows a similar launch on desktop Chrome in the US. The rollout extends Gemini's presence across Google's browser ecosystem, continuing the company's strategy of embedding AI directly into its core consumer products.[140][141]
2025 (November 17) Education initiative Google announces three new Gemini certifications for education: Gemini Certified Educator, focused on teaching innovation; Gemini Certified University Student, aimed at academic and professional readiness; and Gemini Certified Student (K–12), emphasizing responsible AI use. All exams are free, multiple choice, and available in 12 languages, with further localization planned. The initiative reflects Google's strategy to establish Gemini as the default AI platform in educational contexts, competing with Microsoft's AI for Education programs and OpenAI's educational partnerships.[142][143]
2025 (December 12) Update Google announces an upgrade to its Gemini Deep Research agent and opens agent access to developers through a new Interactions API. The API provides a unified interface for Gemini models, built-in agents, and custom agents, supporting extended reasoning and advanced tool use. Powered by Gemini 3 Pro, the Deep Research agent performs long-running, autonomous research tasks, including query formulation, result evaluation, and iterative search. Google reports improved web navigation and reinforcement learning. The agent is available in preview via Google AI Studio.[144]
2026 (January 12) Adoption Apple announces a multiyear partnership with Google to power AI features, including a major Siri upgrade expected in 2026, using Google's Gemini models and cloud technology. Bloomberg reports Apple pays approximately $1 billion per year for the arrangement. The deal is particularly striking given Apple's historically cautious stance toward third-party AI dependencies and its existing integration of OpenAI's ChatGPT for complex queries — Apple reportedly also evaluated proposals from OpenAI and Anthropic before selecting Google. The models process queries through Apple's Private Cloud Compute framework, meaning Google does not receive users' personal data. The agreement signals growing confidence in Google's AI strategy and reflects the broader trend of major technology companies converging on a small number of foundational AI providers, while also deepening Google's already extensive financial relationship with Apple, which separately pays Google tens of billions annually to remain the default search engine on Apple devices.[145][146][147]
2026 (February 3) Notable comment An article characterizes Gemini as a comparatively cautious and procedural AI assistant, reflecting Google’s emphasis on risk management, human oversight, and regulatory compliance. Its interaction style is described as formal and direct, often resembling a tool-like interface rather than a conversational persona. The article notes a past malfunction in which Gemini produced unusually self-critical responses, later corrected by Google. Overall, Gemini is designed to prioritize safety and reliability, with strict limitations on generating sexual content, violent material, self-harm guidance, or politically sensitive misinformation. In contrast to more expressive or provocative models, Gemini represents a conservative design approach focused on minimizing harm, maintaining consistency, and protecting institutional credibility.[148]
2026 (February 19) Update Gemini 3.1 Pro is launched as an upgraded core model in Google's Gemini 3 series, delivering significant gains in advanced reasoning and complex problem-solving. It achieves a verified 77.1% score on the ARC-AGI-2 benchmark, more than doubling the reasoning performance of Gemini 3 Pro. Gemini 3.1 Pro is deployed across consumer, developer, and enterprise products, supporting advanced applications in research, engineering, data synthesis, and creative workflows. The ARC-AGI-2 benchmark, developed by François Chollet, is designed to measure general fluid intelligence rather than task-specific performance, making the score gain notable as an indicator of broader reasoning capability improvements.[149][150]

Visual data

The chart below shows Google Trends data for Bard (chatbot), from January 2023 until December 25, 2023, when the screenshot as taken. Interest is also ranked by country and displayed on world map.[151]

The chart below shows Google Trends data for Gemini (chatbot), from December 2023 to December 2025, when the screenshot as taken. Interest is also ranked by country and displayed on world map.[152]

Wikipedia views

The chart below shows Wikipedia views data for English article Bard (chatbot), from January to November 2023.[153]


The chart below shows Wikipedia views data for English article Gemini (chatbot), from January to December 2025.[154]

Meta information on the timeline

How the timeline was built

The initial version of the timeline was written by Sebastian Sanchez.

Funding information for this timeline is available.

Feedback and comments

Feedback for the timeline can be provided at the following places:

  • FIXME

What the timeline is still missing

Timeline update strategy

Pingbacks

See also

References

  1. "Bard is now Gemini: How to try Ultra 1.0 and the new mobile app". blog.google. Google. 8 February 2024. Retrieved 16 December 2025.
  2. Elias, Jennifer (December 13, 2022). "Google execs warn company's reputation could suffer if it moves too fast on AI-chat technology". CNBC. Archived from the original on December 13, 2022. Retrieved February 6, 2023.
  3. Vincent, James (December 14, 2022). "Google won't launch ChatGPT rival because of 'reputational risk'". The Verge. Archived from the original on December 14, 2022. Retrieved February 6, 2023.
  4. "Towards a Conversational Agent that Can Chat About…Anything". blog.research.google. 28 January 2020. Retrieved 25 December 2023.
  5. "Meena is Google's attempt at making true conversational AI". VentureBeat. 28 January 2020. Retrieved 25 December 2023.
  6. "Google says its new chatbot Meena is the best in the world". MIT Technology Review. 30 January 2020. Retrieved 25 December 2023.
  7. "LaMDA: our breakthrough conversation technology". Google. 18 May 2021. Retrieved 26 December 2023.
  8. "Google I/O 2021: Google unveils LaMDA". ZDNET. Retrieved 26 December 2023.
  9. "Introducing ChatGPT". openai.com. Retrieved 26 December 2023.
  10. Mailk, Aisha (February 3, 2023). "Google tries to reassure investors on AI progress as ChatGPT breathes down its neck". TechCrunch. Archived from the original on February 3, 2023. Retrieved February 6, 2023.
  11. Newton, Casey (May 12, 2023). "How Google is making up for lost time". The Verge. Archived from the original on May 12, 2023. Retrieved June 10, 2023.
  12. Southern, Matt G. (20 April 2023). "Google Bard's Latest Update Boosts Creativity With More Drafts". Search Engine Journal. Retrieved 21 June 2023.
  13. 13.0 13.1 Elias, Jennifer (21 March 2023). "Google CEO tells employees that 80,000 of them helped test Bard A.I., warns 'things will go wrong'". CNBC. Retrieved 18 December 2023.
  14. Patnaik, Davey Alba and Subrat (8 February 2023). "Google suffers $144b wipeout after Bard AI chatbot gives wrong answer". The Sydney Morning Herald. Retrieved 23 May 2023.
  15. Milmo, Dan (9 February 2023). "Google AI chatbot Bard sends shares plummeting after it gives wrong answer". The Guardian. Retrieved 1 December 2023.
  16. Elias, Jennifer (13 February 2023). "Alphabet Chairman John Hennessy explains why Google was hesitant to put out its ChatGPT competitor". CNBC. Retrieved 23 May 2023.
  17. Elias, Jennifer (16 February 2023). "Google asks employees to rewrite Bard's bad responses, says the A.I. 'learns best by example'". CNBC. Retrieved 23 May 2023.
  18. Langley, Hugh. "Google employees are testing a smarter version of its chatbot called 'Big Bard' as it competes with ChatGPT". Business Insider. Retrieved 18 December 2023.
  19. "Google Gemini". Wikipedia. Retrieved 18 December 2023.
  20. Vincent, James (21 March 2023). "Google opens early access to its ChatGPT rival Bard — here are our first impressions". The Verge. Retrieved 24 May 2023.
  21. "Try Bard and share your feedback". Google. 21 March 2023. Retrieved 25 December 2023.
  22. Vincent, James (31 March 2023). "Google CEO Sundar Pichai promises Bard AI chatbot upgrades soon: 'We clearly have more capable models'". The Verge. Retrieved 18 December 2023.
  23. 23.0 23.1 23.2 23.3 23.4 Shewale, Rohit (31 July 2023). "30+ Google Bard Statistics 2023 (Trends & Demographics)". demandsage.com. Retrieved 13 December 2023.
  24. 24.00 24.01 24.02 24.03 24.04 24.05 24.06 24.07 24.08 24.09 24.10 24.11 24.12 24.13 24.14 "‎Bard's Latest AI Capability Updates & Improvements - Bard". bard.google.com. Retrieved 26 December 2023.
  25. Vincent, James (19 April 2023). "Google employees label AI chatbot Bard "worse than useless" and "a pathological liar": report". The Verge. Retrieved 13 September 2023.
  26. "Google's Rush to Win in AI Led to Ethical Lapses, Employees Say". Bloomberg.com. 19 April 2023. Retrieved 13 September 2023.
  27. "Code and debug with Bard". Google. 21 April 2023. Retrieved 25 December 2023.
  28. "Google Bard is now open for Workplace accounts in its latest update". Neowin. 13 December 2023. Retrieved 13 December 2023.
  29. "Google Bard AI is now available for Workspace accounts". 9to5google.com. Retrieved 13 December 2023.
  30. "Google's Bard AI tool now available for Workspace accounts as well". Hindustan Times. 6 May 2023. Retrieved 13 December 2023.
  31. Staff, Beebom (6 May 2023). "Google Workspace Accounts Finally Gain Access To Google Bard". Beebom. Retrieved 13 December 2023.
  32. "What's ahead for Bard: More global, more visual, more integrated". Google. 10 May 2023. Retrieved 24 May 2023.
  33. Miller, Ron (10 May 2023). "Google ends Bard waitlist, making English version of chatbot widely available". TechCrunch. Retrieved 21 June 2023.
  34. Southern, Matt G. (10 May 2023). "Google Bard Removes Waitlist, Adds Image & Coding Features". Search Engine Journal. Retrieved 23 May 2023.
  35. Lardinois, Frederic (10 May 2023). "Google launches a smarter Bard". TechCrunch. Retrieved 23 May 2023.
  36. Schoon, Ben (15 May 2023). "Latest Google Bard update improves summaries and sourcing". 9to5Google. Retrieved 26 December 2023.
  37. "Google Bard's new update improves summaries, sourcing". IMDb. Retrieved 26 December 2023.
  38. Goodwin, Danny (23 May 2023). "Google Bard adds images in responses". Search Engine Land. Retrieved 24 May 2023.
  39. "Google rolls out 'Magic Compose', powered by Bard. Here's what this tool does". Hindustan Times. 29 May 2023. Retrieved 1 June 2023.
  40. "ChatGPT vs Bing Chat vs Google Bard: Which is the best AI chatbot?". ZDNET. Retrieved 1 June 2023.
  41. "Bing CEO: Google Bard Uses A Much Smaller Model Than Bing Chat". seroundtable.com. Retrieved 1 June 2023.
  42. "ChatGPT vs Bing Chat vs Google Bard: Which is the best AI chatbot?". ZDNET. Retrieved 1 June 2023.
  43. "Google Bard can now deliver results based on your precise location - if you want". ZDNET. Retrieved 21 June 2023.
  44. "Google Bard is much better at solving coding and mathematical questions now". XDA Developers. 8 June 2023. Retrieved 21 June 2023.
  45. "Bard is getting better at logic and reasoning". Google. 7 June 2023. Retrieved 25 December 2023.
  46. "Top 5 Benefits of Using Google Bard AI". Analytics Insight. 9 June 2023. Retrieved 21 June 2023.
  47. "Google Bard gets smarter with improved math and data analysing ability". The Indian Express. 8 June 2023. Retrieved 13 September 2023.
  48. "Irish data watchdog blocks Google from launching Bard in the EU: Report". Cointelegraph. 13 June 2023. Retrieved 21 June 2023.
  49. "Google Bard Gets Lens, Lens Detects Skin Conditions & Helps You Shop". seroundtable.com. Retrieved 21 June 2023.
  50. Quach, Katyanna. "Google doesn't want its employees using Bard code". www.theregister.com. Retrieved 21 June 2023.
  51. Dean, Grace. "A lawsuit claims Google has been 'secretly stealing everything ever created and shared on the internet by hundreds of millions of Americans' to train its AI". Business Insider. Retrieved 13 July 2023.
  52. Thorbecke, Catherine (11 July 2023). "Google hit with lawsuit alleging it stole data from millions of users to train its AI tools". CNN. Retrieved 21 December 2023.
  53. "Google launches ChatGPT rival Bard in Europe and Brazil". France 24. 13 July 2023. Retrieved 14 July 2023.
  54. "Google's Bard AI Chatbot Now Supports Over 40 Languages". CNET. Retrieved 13 September 2023.
  55. "Bard's latest update: more features, languages and countries". Google. 13 July 2023. Retrieved 25 December 2023.
  56. "Google Bard outperforms ChatGPT with these 5 incredible use cases". indianexpress.com. Retrieved 13 September 2023.
  57. PM, Aiswarya (26 July 2023). "5 Reasons Google Bard is Better Than ChatGPT". Analytics Insight. Retrieved 13 September 2023.
  58. "10 helpful ways to use Bard". Google. 10 August 2023. Retrieved 25 December 2023.
  59. "Google Bard's Latest Update Boosts Creativity With More Drafts". Search Engine Journal. 20 April 2023. Retrieved 25 December 2023.
  60. "OpenAI ChatGPT, Google Bard spreading news-related misinformation: Report". ETTelecom.com. Retrieved 13 September 2023.
  61. "Google Bard misinforms users 8 out of 10 times without disclaimers". Fortune. Retrieved 13 September 2023.
  62. "A Bard's Tale – how fake AI bots try to install malware". www.welivesecurity.com. Retrieved 13 September 2023.
  63. "Google sues alleged operators of malware scheme that used Bard AI as lure". Reuters. Retrieved 13 September 2023.
  64. "Google Links Bard AI Tool to Its Other Products". VOA. 20 September 2023. Retrieved 22 September 2023.
  65. "Bard can now connect to your Google apps and services". Google. 19 September 2023. Retrieved 25 December 2023.
  66. "Assistant with Bard: A step toward a more personal assistant". Google. 4 October 2023. Retrieved 24 December 2023.
  67. Knight, Will. "Google Assistant Finally Gets a Generative AI Glow-Up". Wired. Retrieved 24 December 2023.
  68. "Google to combine generative AI chatbot with virtual assistant". reuters.com. Retrieved 24 December 2023.
  69. Roth, Emma (4 October 2023). "Google is launching a generative AI-enhanced version of Assistant". The Verge. Retrieved 11 October 2023.
  70. Roth, Emma (7 November 2023). "Google contractors objected to reading obscene Bard prompts — now they're unionizing". The Verge. Retrieved 18 December 2023.
  71. "Group of Google contractors who work on Search and Bard win union vote". CNBC. Retrieved 18 December 2023.
  72. "Bard AI enhances YouTube content interpretation in new Google update". readwrite.com. Retrieved 1 December 2023.
  73. "Google Has Enhanced Bard's Understanding Of YouTube Videos". forbes.com. Retrieved 1 December 2023.
  74. "You can now ask questions about YouTube videos to Google's Bard AI". innovation-village.com. Retrieved 1 December 2023.
  75. Pierce, David (6 December 2023). "Google launches Gemini, the AI model it hopes will take down GPT-4". The Verge. Retrieved 22 December 2023.
  76. "Google DeepMind Unveils Its Most Powerful AI Offering Yet". TIME. 6 December 2023. Retrieved 22 December 2023.
  77. "Google launches Gemini, upping the stakes in the global AI race". AP News. 6 December 2023. Retrieved 22 December 2023.
  78. "Colin Murdoch from Google DeepMind: 'Gemini will transform the way billions of people live and work'". El País. Retrieved 12 August 2024.
  79. "Google Gemini". Wikipedia. Retrieved 12 August 2024.
  80. "Samsung's Galaxy S24 Will Feature Google Gemini-Powered AI Features". techcrunch.com. Retrieved 15 December 2024.
  81. "Google Launches Gemini-Powered Search Ads to More Advertisers". searchenginejournal.com. Retrieved 15 December 2024.
  82. "Google's Gemini Pro Beats GPT-4". analyticsindiamag.com. Retrieved 23 December 2024.
  83. "Chatbot Arena Leaderboard". HuggingFace. Retrieved 23 December 2024.
  84. "Google's Bard Chatbot Gets the Gemini Pro Update Globally". TechCrunch. Retrieved 12 August 2024.
  85. "Google turns Bard AI into Gemini, launches Gemini Advanced". techtarget.com. Retrieved 23 December 2024.
  86. "Why Google Renamed Bard to Gemini". cmswire.com. Retrieved 23 December 2024.
  87. "Bard to Gemini Advanced app". blog.google. Retrieved 30 December 2024.
  88. "Google Launches Its New Version of AI Tool Gemini". the-tartan.org. Retrieved 30 December 2024.
  89. "Google Gemini: Next Generation Model". blog.google. Retrieved 30 December 2024.
  90. "Google AI Gemini criticized for inaccuracies in historical content". theverge.com. Retrieved 23 December 2024.
  91. "Google pauses AI tool Gemini's ability to generate images of people after historical inaccuracies". TechCrunch. 22 February 2024. Retrieved 23 December 2024.
  92. "Google Gemini Bugs Enable Prompt Leaks, Injection via Workspace Plugin". SC World. Retrieved 2024-02-23.
  93. "AI: Google restricts Gemini chatbot election answers". BBC News. Retrieved 2024-02-23.
  94. "Google's Gemini Still Restricts Political Answers in 2024". AI Chief. Retrieved 27 January 2026.
  95. Andrew Mihalache, Justin Grad, Nikhil S. Patil, Ryan S. Huang, Marko M. Popovic, Ashwin Mallipatna, Peter J. Kertes, Rajeev H. Muni (2024). "Google Gemini and Bard artificial intelligence chatbot performance in ophthalmology knowledge assessment". Eye. 38: 2530–2535. doi:10.1038/s41433-024-03067-4. Retrieved 2024-02-23.{{cite journal}}: CS1 maint: multiple names: authors list (link)
  96. "AI chatbots in medicine: performance benchmarks and limitations". Nature/Eye. Retrieved 2024-02-23.
  97. "Google Declares It the 'Gemini Era' as Revenue Grows 15%". Search Engine Journal. 11 January 2025. Retrieved 13 January 2025.
  98. "Alphabet Q1 2024 Earnings Call Transcript". Alphabet Investor Relations. April 25, 2024. Retrieved 13 January 2025.
  99. "Google Gemini Update: May 2024". Google Blog. Retrieved 3 November 2024.
  100. "Google Launches Gemini Nano for Chrome Desktop Client". Okoone. Retrieved 3 November 2024.
  101. 101.0 101.1 "Google faces lawsuit over Gemini AI". theregister.com. Retrieved 30 December 2024.
  102. "Exploring the future of media: Insights on AI and technology". Arab News. Retrieved 3 November 2024.
  103. "Arabic users get Gemini AI upgrade with new app and advanced mode". Edge Middle East. Retrieved 3 November 2024.
  104. "Google Gemini app now available in the EU and UK for Android users". Testing Catalog. Retrieved 3 November 2024.
  105. "Google opens access to 2 million context window of Gemini 1.5 Pro". InfoWorld. Retrieved 3 November 2024.
  106. "NBC partners with Google's Gemini AI for enhanced Paris Olympics coverage". Quartz. Retrieved 3 November 2024.
  107. "Google, Team USA and NBCUniversal Strike New Partnership". NBCUniversal. July 18, 2024. Retrieved 3 November 2024.
  108. "Google AI Features Coming to NBC 2024 Paris Olympics Coverage". The Hollywood Reporter. July 18, 2024. Retrieved 3 November 2024.
  109. "Google's free Gemini chatbot gets 1.5 flash update, making responses faster and smarter". SiliconANGLE. Retrieved 3 November 2024.
  110. "Google Gemini: New Features in July 2024". Google Blog. Retrieved 3 November 2024.
  111. "AI Model for Cough Disease Detection". blog.google. Retrieved 3 November 2024.
  112. "Google's Health Acoustic Representations (HeAR)". Google Research. Retrieved 3 November 2024.
  113. "NotebookLM: Audio Overviews". blog.google. Retrieved 3 November 2024.
  114. "Gemini Live Extensions and Language Expansion". blog.google. Retrieved 3 November 2024.
  115. "Google Maps Gets Gemini AI to Answer Travel, Weather, and Other Questions". androidheadlines.com. Retrieved 3 November 2024.
  116. "Gemini and Google Maps Navigation Updates". blog.google. Retrieved 3 November 2024.
  117. "Google's AI Chatbot Gemini Tells User to 'Please Die' and 'You Are a Waste of Time and Resources'". Sky News. Retrieved 19 November 2024.
  118. "Gemini under fire after telling user to 'please die' — here's Google's response". Tom's Guide. Retrieved 19 November 2024.
  119. "Android Gains Gemini AI Integration for Enhanced User Experience". InfoQ. Retrieved 19 November 2024.
  120. "Google's AI Reasoning Model Gemini Flash 2 Introduced". theverge.com. Retrieved 15 December 2024.
  121. "North Korean hackers use Google's Gemini AI for cybercrime and espionage". NK News. January 2025. Retrieved March 22, 2025.
  122. "Hackers use Google Gemini to aid in cyberattacks, Google's own security team finds". Fortune. January 30, 2025. Retrieved March 22, 2025.
  123. "How government hackers are trying to exploit Google Gemini AI". Computer Weekly. January 29, 2025. Retrieved March 22, 2025.
  124. "Google launches free Gemini Code Assist tier for individuals". SiliconANGLE. February 25, 2025. Retrieved March 22, 2025.
  125. "Google officially launches Gemini with personalization". GSMArena. March 21, 2025. Retrieved March 22, 2025.
  126. Waters, John K. (June 25, 2025). "Google unveils Gemini CLI for developers". InfoWorld. Retrieved July 17, 2025.
  127. "Gemini CLI: Your open-source AI agent". Google Blog. June 25, 2025. Retrieved July 17, 2025.
  128. Schwartz, Eric Hal (July 9, 2025). "Google just announced 5 new Gemini features coming to Android". TechRadar. Retrieved July 17, 2025.
  129. "What's new in Android 16". Google Blog. July 9, 2025. Retrieved July 17, 2025.
  130. Makwana, Divyang (July 17, 2025). "Google launches Gemini 2.5 Pro, Deep Search, and AI‑Powered Business Calling". MobiGyaan. Retrieved July 17, 2025.
  131. "Google rolls out AI-powered business-calling feature". TechCrunch. July 16, 2025. Retrieved July 17, 2025.
  132. Southern, Matt G. (August 13, 2025). "Google Gemini Adds Personalization From Past Chats". Search Engine Journal. Retrieved October 15, 2025.
  133. "Gemini Apps Privacy Hub". Google Support. Retrieved October 15, 2025.
  134. Balkan Green Energy News (August 28, 2025). [[1](https://balkangreenenergynews.com/google-reveals-gemini-ais-energy-and-water-consumption/) "Google reveals Gemini AI's energy and water consumption"]. Balkan Green Energy News. Retrieved October 14, 2025. {{cite web}}: Check |url= value (help)
  135. Perez, Sarah (2025-09-05). "Google Gemini dubbed 'high risk' for kids and teens in new safety assessment". TechCrunch. Retrieved 2025-10-14.
  136. "Common Sense Media AI ratings". Common Sense Media. Retrieved 2025-10-14.
  137. "Gemini achieves gold-medal performance at ICPC". Google Blog. Retrieved October 14, 2025.
  138. Fortin, Raisinghani, Odoom, Vernade, Alisa, Naina, Seth, Guillaume (October 2, 2025). "Gemini 2.5 Flash Image now ready for production with new aspect ratios". Google Developers Blog. Retrieved October 14, 2025.{{cite web}}: CS1 maint: multiple names: authors list (link)
  139. Pichai, Sundar (October 9, 2025). "Gemini Enterprise: The new front door for Google AI in your workplace". Google Blog. Retrieved October 14, 2025.
  140. Li, Abner (October 13, 2025). "Gemini overlay rolls out 'Summarize page' shortcut in Chrome for Android". 9to5Google. Retrieved October 14, 2025.
  141. "Chrome for Android release notes". Chromium. Retrieved October 14, 2025.
  142. "Announcing new Gemini certifications for education". Google Blog. Google. Retrieved 13 December 2025.
  143. "Google launches AI certifications for teachers and students". Education Week. Retrieved 13 December 2025.
  144. "Google Rolls Out Upgraded Gemini Deep Research And Opens Agent Access To Developers". Nasdaq. Retrieved 13 December 2025.
  145. "Apple picks Google's Gemini to run AI-powered Siri coming this year". CNBC. CNBC LLC. January 12, 2026. Retrieved February 20, 2026.
  146. "Google's Gemini to power Apple's AI features like Siri". TechCrunch. January 12, 2026. Retrieved February 20, 2026.
  147. "Apple teams up with Google Gemini for AI-powered Siri". CNN. January 12, 2026. Retrieved February 20, 2026.
  148. Booth, Robert (3 February 2026). "From 'nerdy' Gemini to 'edgy' Grok: how developers are shaping AI behaviours". The Guardian. Retrieved 18 February 2026.
  149. "Gemini 3.1 Pro: A smarter model for your most complex tasks". Google Blog. Google. 19 February 2026. Retrieved 20 February 2026.
  150. "ARC-AGI-2 benchmark results". ARC Prize. Retrieved 20 February 2026.
  151. "Google Trends". Google Trends. Retrieved 26 December 2023.
  152. "Google Trends – "Gemini (chatbot)"". Google Trends. Retrieved 16 December 2025.
  153. "Wikipedia Views: results". wikipediaviews.org. Retrieved 26 December 2023.
  154. "Gemini (chatbot) - Wikipedia Views". wikipediaviews.org. Retrieved 16 December 2025.