Timeline of AI in writing

This is a Timeline of AI in writing, covering the development of artificial intelligence tools for generating, editing, and analyzing written text, from early computational style checkers at Bell Labs in the 1970s through the large language models and generative AI platforms that reshape writing practices today.

Sample questions

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

• When did grammar checking move from Unix mainframes to personal computers, and what were the key milestones along the way?
• What was the first major newspaper to use AI to automatically generate articles, and at what scale?
• Which AI-generated novel won a prestigious literary prize before its author disclosed the AI's role, and how did the literary world respond?
• What architecture underlies GPT, BERT, and virtually all modern AI writing tools, and when was it introduced?
• How did the market for standalone grammar-checking software end, and what replaced it?
• What legal actions have authors taken against AI companies over the use of their work in training data?
• When did a major peer-reviewed journal editorial first declare that human and machine writing had become indistinguishable?
• Which AI writing tool was the first to bring grammar assistance to a mass-market word processor, and which researchers built it?
• How have professional writers — novelists, poets, journalists, PR professionals — publicly responded to the rise of generative AI in their fields?
• What is the difference between natural language generation (NLG) tools of the 2010s and the large language models that followed, and which writing domains did each serve?

Big picture

Time period	Development summary	More details

Full timeline

Inclusion criteria

The timeline includes events that directly bear on the use, development, or cultural reception of artificial intelligence in written communication. Specifically, a row qualifies if it falls into one or more of the following categories:

• Technical milestones: development or release of models, architectures, tools, or software systems that materially advance AI's capacity to generate, edit, analyze, or assist with written text.
• Product launches and integrations: commercial deployment of AI writing tools to a broad user base, including embedding of AI writing features into existing platforms.
• Legal and policy events: lawsuits, regulatory actions, or institutional guidelines that directly concern AI-generated text, authorship, or copyright.
• Research publications: peer-reviewed or widely cited studies that advance understanding of how AI affects writing practice, quality, or education.
• Commentary and advocacy: opinion pieces, open letters, or public statements that reflect significant turning points in how writers, educators, or institutions perceive AI's role in writing — included selectively, prioritizing pieces that generated broader discussion or represent a major position in an ongoing debate rather than individual opinions.
• Newsroom and institutional adoption: documented cases of organizations adopting AI for writing-related workflows, prioritizing cases with broader sectoral significance.

Rows concerning AI speech, audio, or image generation are excluded unless they directly intersect with written language. Individual blog posts or opinion pieces are included only when they represent a significant moment in public or professional discourse, not merely because they address the topic.

Year	Writing domain	Event type	Event description
Pre-1966	Writing tools and technology	Backdrop	Throughout the first half of the twentieth century, professional writing is produced on manual and electric typewriters, with revision requiring physical cutting, pasting, and complete retyping of pages. The introduction of the IBM MT/ST in 1964 — the first machine IBM markets as a word processor — begins to change this, allowing typed text to be stored on magnetic tape and edited before printing without retyping entire pages. By the mid-1970s, dedicated word processing systems such as the Wang WPS (1976) are in widespread corporate use, processing written text mechanically but offering no analysis, feedback, or generation of language. The entire burden of grammar, style, clarity, and correctness rests with the human writer. It is in this context — mechanical text production with no computational intelligence applied to the language itself — that the first AI writing analysis tools of the late 1960s and 1970s emerge, beginning with ELIZA (1966) and the Bell Labs programs developed by Lorinda Cherry (~1976). These early systems do not generate text or check grammar in any commercial sense; they analyze and comment on text already written, as a critic or editor might. The subsequent decades see a progressive shift: from analysis-only tools to grammar and style checkers, from standalone desktop programs to embedded word processor features, and ultimately to cloud-based and generative systems that write alongside — or instead of — their human users.
1966	Dialogue systems	Tool development	Joseph Weizenbaum develops *ELIZA*, an early rule-based chatbot that mimics human-like conversation using pattern-matching techniques. While simple, it marks the first instance of AI simulating natural written interaction.
1971 (February)	Style and grammar analysis	Tool development	Ralph Gorin, a graduate student at Stanford University's Artificial Intelligence Laboratory working under computer scientist Les Earnest, creates SPELL for the DEC PDP-10 — the first interactive spell-checking program written as a general-purpose application for English text rather than as a research prototype. Written in assembly language for speed, SPELL searches a word list for plausible corrections differing by a single letter or adjacent transposition and presents suggestions to the user interactively, making it the first spelling corrector as well as the first spell checker. Gorin makes SPELL publicly accessible following standard practice at the Stanford AI Lab, and it spreads rapidly through the nascent ARPANET, reaching universities and research institutions worldwide. By the late 1970s, spell checkers based on SPELL's approach become standard tools on mainframe computers at universities and large corporations, though their use requires expensive mainframe access unavailable to most writers.
1976	Technical writing and editing	Product integration	The European Commission licenses SYSTRAN, a rule-based machine translation system developed by Hungarian-American linguist Peter Toma and founded in San Diego in 1968, initially to translate documents between English and French. Work begins on a new English-to-French system in 1976, with French-to-English following the next year and a third language combination in 1979. Originally developed for the United States Air Force to translate Russian scientific and technical documents during the Cold War, SYSTRAN becomes under the Commission the first AI system deployed at institutional scale to produce written output in a professional context — generating translated prose from structured source documents for use across EU member states. The adoption establishes post-editing as a professional writing workflow: human translators review and correct machine-generated text rather than producing prose from scratch, prefiguring the human-AI collaborative writing practices that become widespread with large language models decades later. By 1982, SYSTRAN handles 50% of the Commission's English-to-Italian translation workload.
~1976	Style and grammar analysis	Tool development	American computer scientist Lorinda Cherry at Bell Labs in Murray Hill, NJ, begins developing programs to analyze English texts for weaknesses in diction and style.
Late 1970s	Technical writing and editing	Tool development	Bell Labs’ Documentation Technologies Group in Piscataway, NJ, creates complementary programs forming the UNIX™ Writer’s Workbench Software suite.
1980	Style and grammar analysis	Product launch	WordCheck, developed for Commodore personal computer systems, becomes the first commercial spell-checking software available for personal computers, marking the transition of spell checking from an exclusively mainframe-based research tool to a product accessible to ordinary writers. Where SPELL and its successors had required access to expensive institutional computing infrastructure, WordCheck brings automated spelling correction to desktop machines costing a fraction of a mainframe, prefiguring the integration of spell checking into mass-market word processors that follows in the early 1980s.
1981	Style and grammar analysis	Tool development	Aspen Software of Tijeras, New Mexico, releases Grammatik, the first grammar-checking program for personal computers. Developed by computer scientist Bruce Wampler, who had found Writer's Workbench invaluable during his doctoral work but was frustrated by having to move between a personal computer and a Unix mainframe to use it, Grammatik brings automated writing analysis to individual desktop users for the first time. Initially available for the TRS-80, it soon ports to CP/M and the IBM PC. Reference Software International acquires Grammatik in 1985 and develops it into a full grammar checker capable of detecting errors beyond diction and style. WordPerfect Corporation later integrates it directly into its word processor, completing the arc from Unix research tool to mass-market writing aid.
1982	Style and grammar analysis	Research development	IBM researchers George Heidorn, Karen Jensen, Lance Miller, Roy Byrd, and Martin Chodorow publish a description of EPISTLE, an experimental text-critiquing system designed for business correspondence, in the IBM Systems Journal. Unlike earlier pattern-matching tools such as Writer's Workbench and Grammatik, EPISTLE applies full syntactic parsing to sentences using a computerized English grammar, operating at three levels: word lookup, grammar processing, and style analysis. The system achieves a 64% parse success rate on business letters tested as of December 1981. It marks what one later analysis calls the first "really visible foray" into second-generation grammar checking based on real syntactic processing. Heidorn and Jensen later move to Microsoft, where they lead development of the grammar checker built into Microsoft Word.
1992	Style and grammar analysis	Product integration	Microsoft integrates grammar checking into Microsoft Word by licensing CorrecText, a pattern-based checker developed by Houghton Mifflin, making grammar assistance a built-in feature of a mass-market word processor for the first time. WordPerfect Corporation responds by acquiring Reference Software International and bundling Grammatik directly into WordPerfect. The simultaneous move by both dominant word processors effectively ends the market for standalone grammar-checker products. The CorrecText-based checker is a first-generation pattern-matching tool; a more sophisticated full-parsing grammar engine, developed by former IBM Epistle researchers George Heidorn and Karen Jensen, would not reach Word users until 1997, when Microsoft replaces CorrecText with their parser-based system.
1997	Style and grammar analysis	Product integration	Microsoft releases Word 97, introducing a grammar checker built on a full natural language processing system developed by George Heidorn and Karen Jensen — the same researchers who had built IBM's EPISTLE system fifteen years earlier. Unlike the pattern-matching CorrecText engine it replaces, the new checker parses sentences syntactically, enabling it to detect a far wider range of grammatical errors. One later analysis describes the upgrade as "a major breakthrough" in automated writing assistance, and notes that the full-parsing Word grammar checker remains for years the only publicly well-documented commercial-grade syntax checker at scale. The Heidorn–Jensen system would remain the core of Word's grammar checking for over two decades, until Microsoft replaces it with a neural network–based checker in the 2020s.
2006	General-purpose writing and content creation	Product launch	Richard Rosenblatt and Shawn Colo found Demand Media in early 2006, acquiring eHow and building a vertically integrated content operation that becomes the defining example of the content farm model. Using proprietary algorithmic tools to identify high-value search queries and then commissioning freelance writers to produce articles targeting those queries at minimum cost — typically $15–$20 per article — Demand Media adds around 5,000 new articles and videos to eHow daily at its peak. The AI component of the operation lies in the demand-side optimization: algorithms determine what to write, while human writers produce the text, making content farms an early example of algorithmically directed mass writing production. Similar operations emerge across the web, including Associated Content, Suite101, and Examiner.com. The model provokes widespread criticism for degrading the quality of written information available online and crowding out substantive content in search results. In February 2011, Google releases its Panda (Google) algorithm update — informally called the "farmer update" — specifically targeting thin, low-quality content, reducing traffic to Demand Media sites by up to 40% and effectively ending the first era of algorithmically driven mass writing production. The episode establishes a template for the later debate about AI-generated content: scale and optimization against quality and human judgment, with search engine intervention as the enforcement mechanism.
2009 (July)	Style and grammar analysis	Product launch	Ukrainian software developers Alex Shevchenko, Max Lytvyn, and Dmytro Lider launch Grammarly initially as a grammar and plagiarism checker aimed at university students, building on an earlier plagiarism-detection product called MyDropbox that the founders had developed and sold to Blackboard Inc.. Where grammar checking had previously required either a standalone desktop application or a built-in word processor feature, Grammarly operates as a cloud-based service accessible through browser extensions, making AI writing assistance available across any text field on any website. The company later evolves into a broader AI-powered writing platform offering style, tone, and clarity suggestions, reaching a reported valuation of $13 billion by 2021. Its architecture — real-time, cross-platform, cloud-based — becomes the model for a subsequent generation of AI writing assistants.
2010	Journalism and technical writing	Company founded	Northwestern University computer science professors Kristian Hammond and Larry Birnbaum co-found Narrative Science in Evanston, Illinois, commercializing a student research project called StatsMonkey that automatically generated college baseball game stories from box scores and play-by-play data. The company develops Quill, a natural language generation platform that extracts the most significant patterns from structured datasets and converts them into readable narrative prose. Quill's early clients include financial institutions and media organizations seeking to cover data-rich domains — earnings reports, sports results, market summaries — at a scale impossible with human writers alone. The company is later acquired by Salesforce in 2021. Narrative Science and its contemporary Automated Insights establish NLG as a commercially viable writing technology in the years before large language models render the underlying rule-based approach largely obsolete.
2014 (July)	Journalism and technical writing	Newsroom automation	The Associated Press begins using Wordsmith, a natural language generation platform developed by Automated Insights of Durham, North Carolina, to automatically produce corporate earnings reports from structured financial data supplied by Zacks Investment Research. Where AP reporters previously produced around 300 such stories per quarter, Wordsmith generates over 3,000 in the same period — a tenfold increase — freeing journalists to focus on investigative and enterprise reporting. The project demonstrates that NLG can produce publishable prose at scale for data-rich domains and prompts widespread discussion in newsrooms about the future division of labor between human writers and automated systems. AP later expands the program to cover minor league baseball and announces plans to extend coverage to Canadian and European company earnings.
2017 (June 12)	Text generation and comprehension	Research publication	Eight researchers at Google Brain and Google Research — Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin — publish "Attention Is All You Need", introducing the Transformer architecture for sequence-to-sequence tasks. Where dominant language models at the time relied on recurrent or convolutional neural networks that processed text sequentially, the Transformer dispenses with recurrence entirely in favor of a self-attention mechanism that relates all positions in a sequence simultaneously, making training substantially more parallelizable. Originally developed for machine translation, the architecture proves broadly applicable to any task involving sequential text. It becomes the foundation of BERT (2018), GPT-2 (2019), and the entire subsequent lineage of large language models that reshape AI-assisted writing — making this paper arguably the single most consequential technical development in the history of AI writing tools. "Attention Is All You Need" is among the most cited papers in machine learning.
2018 (July 24)	Predictive typing	Product launch	Google announces the rollout of Smart Compose. Initially an experimental Gmail feature, Smart Compose helps autocomplete email sentences, especially for common phrases, greetings, and addresses. It learns users' writing styles over time, including names and jargon, to offer personalized suggestions. Users can accept suggestions by pressing the tab key. The move marks a shift from Google's usual practice of delaying feature releases after announcement. The feature aims to enhance email productivity without replacing full message composition. It becomes a key part of Gmail’s intelligent, AI-driven tools.
2019 (February 18)	Journalism and technical writing	Product launch	OpenAI releases GPT-2, a large language model capable of generating highly coherent and human-like text. While the research is widely praised for its technical achievements, OpenAI’s decision not to release the model’s largest version sparks controversy over ethical concerns. The company cites the potential for misuse in disinformation and abuse at scale. Critics accuse OpenAI of hype-generation.
2021 (June 16)	Creative writing	Product launch	Delaware-based studio Anlatan launches NovelAI, a subscription-based AI writing platform built specifically for fiction, founded by developers from the AI Dungeon community who sought a privacy-focused alternative following controversial policy changes and data breaches at other AI fiction platforms. Unlike general-purpose language models, NovelAI uses fine-tuned versions of GPT-Neo and GPT-J trained specifically on literary text rather than general web data, enabling it to maintain narrative voice, genre conventions, and character consistency more reliably than general AI tools. The platform reaches 40,000 subscribers within months of launch, positioning uncensored creative freedom, user privacy, and encrypted story storage as its defining values. NovelAI's rapid adoption demonstrates significant unmet demand for AI writing assistance among fiction writers, particularly in genre communities — fantasy, science fiction, horror, and romance — where general-purpose tools consistently fail to match the conventions of the form.
2021 (November)	Creative writing	Product launch	Sudowrite, founded in August 2020 by science fiction writers Amit Gupta — a Y Combinator alumnus and serial entrepreneur — and James Yu — an engineer and author who previously worked at Google and Zynga — exits closed beta and raises $3 million in seed funding, backed by the founders of Medium, Twitter, WordPress, and Gumroad alongside writers and directors of major Hollywood productions. Built on GPT-3 and designed explicitly around the cognitive demands of long-form narrative writing — maintaining character voice, managing plot consistency, generating descriptive prose, and overcoming writer's block — Sudowrite is validated early by prominent science fiction and fantasy authors including Cory Doctorow, Mary Robinette Kowal, and Ted Chiang. Unlike NovelAI's uncensored, community-oriented approach, Sudowrite positions itself as a professional creative collaborator that augments rather than replaces the author's voice, later developing a proprietary fiction-trained model called Muse built on data licensed with author consent. The New Yorker later calls it "a salvation" for writers. The platform's emergence alongside NovelAI establishes fiction-specific AI writing assistance as a distinct product category, separate from general-purpose tools, with its own design principles, user communities, and debates about authorship and craft.
2022 (March 4)	Technical writing and task-oriented content	Research publication	OpenAI researchers Long Ouyang, Jeff Wu, Xu Jiang, and colleagues publish "Training language models to follow instructions with human feedback", introducing InstructGPT — a family of models fine-tuned from GPT-3 using reinforcement learning from human feedback (RLHF) to follow user instructions more reliably than the base model. Where GPT-3 generates text by predicting likely continuations of a prompt, InstructGPT is explicitly optimized to produce outputs that match human intent, making it substantially better suited for task-oriented writing applications such as drafting, summarization, and structured content generation. In human evaluations, outputs from the 1.3B parameter InstructGPT model are preferred to those from the 175B GPT-3 despite having 100 times fewer parameters. InstructGPT also generates truthful answers roughly twice as often as GPT-3 and produces 25% fewer toxic outputs. The model directly prefigures ChatGPT, which applies the same RLHF approach to a conversational format and launches eight months later, and establishes instruction-following as the dominant paradigm for deploying language models in writing-assistance contexts.
2022 (March)	Local journalism and weather reporting	Newsroom automation	The Argentine local paper Diario Huarpe starts using AI-powered automation from United Robots to publish around 250 football articles and 3,000 weather reports monthly. This helps its small team expand coverage despite limited staff, especially for sports and local weather. The system uses structured data and Natural Language Generation (NLG) to create articles in natural regional Spanish. While some initial resistance exists, journalists start using the automation to focus on deeper stories. The main challenge is accessing structured data—especially for local leagues—though weather data is readily available. Automation boosts efficiency, SEO traffic, and audience reach in San Juan.
2022 (November 17)	Academic and scientific writing	Product launch	Meta AI introduces Galactica, a language model capable of generating scientific and academic papers from simple text inputs. Trained on a vast corpus of scientific literature, knowledge bases, and reference materials, Galactica compresses this data into a 120-billion parameter model. It aims to summarize academic literature, solve math problems, and generate Wiki articles. However, after its launch, Galactica faces criticism for generating content that sounds grammatically correct but is scientifically inaccurate, leading Meta to pull it down after just three days. Some experts find it useful, while others consider it a "random bullshit generator."
2022 (November 30)	General-purpose writing and content creation	Product launch	OpenAI introduces ChatGPT, a language model designed to hold natural conversations. This launch significantly impacts writing by making advanced language generation accessible to a broad audience. Writers gain a powerful tool for drafting, editing, brainstorming, and generating content across genres and formats. ChatGPT reduces time spent on routine tasks, such as grammar correction or summarization, and helps overcome writer's block by suggesting ideas or alternative phrasings. It also democratizes writing support, benefiting non-native speakers and professionals alike. However, it raises questions about originality, authorship, and reliance on AI-generated text.
2023 (January 3)	Academic and educational writing	Product launch	Edward Tian, a computer science undergraduate at Princeton University building his senior thesis on AI detection, releases the beta version of GPTZero on Twitter during his winter break, describing it as a tool to detect whether text is written by ChatGPT or a human. The tool crashes within its first week from 30,000 users, attracting coverage in the New York Times, Washington Post, and Wall Street Journal. GPTZero uses two metrics — perplexity (randomness of word choice within a sentence) and burstiness (variation of perplexity across sentences) — to distinguish AI from human prose. The launch triggers a wave of competing detection tools, including Turnitin's AI detector (April 2023) and OpenAI's own Text Classifier (January 2023), which OpenAI shuts down just seven months later after widespread criticism of its poor accuracy. Critics raise significant concerns across the detection ecosystem: false positive rates that incorrectly flag human writing as AI-generated, with research showing disproportionate impact on non-native English speakers whose prose patterns more closely resemble AI output. The proliferation of detection tools — and their documented unreliability — deepens the crisis of authenticity in academic writing, intensifying debates about trust, authorship, and the limits of automated judgment already under way in educational institutions.
2023 (c.February 14)	Business and professional writing	Product launch	Full-stack generative AI platform Writer launches Palmyra, a trio of LLMs that focus on business writing and marketing data. The models include Palmyra Small (128M), Palmyra Base (5B), and Palmyra Large (20B), and are aimed at enterprises looking to invest in generative AI. Palmyra LLMs offer both an application layer and a foundation model layer, making Writer the first to provide both on a single platform. The models also offer high levels of security and privacy features. While general-use LLMs can achieve human-like output, they lack contextual awareness, multi-modal inputs, brand integrity and compliance with security and privacy standards, limiting their usefulness for enterprise organizations.
2023 (February)	Professional and creative writing	Commentary	Ken Scudder, a communications trainer and consultant and a regular contributor to the Public Relations Society of America's trade publication Strategies & Tactics, explores the rise of AI-generated text and its implications for professional writers. Using ChatGPT to generate an article on AI and writing, he observes the machine's competent yet generic output — fluent but lacking creativity, personal voice, and lived insight. Scudder argues that writers who embrace AI as a productivity tool rather than fearing it as a replacement will stay relevant, and that the distinctively human qualities of writing — humor, specificity, emotional resonance — remain beyond AI's reach. Writing in a publication read by public relations and communications professionals, his framing represents an early industry-level attempt to define the human writer's competitive advantage in an AI-assisted workplace. Critics of this view, however, argue that the qualities Scudder identifies as uniquely human are themselves subject to erosion as models improve, and that framing AI as merely a tool understates the structural pressure it places on the demand for professional writers.
2023 (March 7)	Academic and scientific writing	Commentary	An editorial in Nature Biomedical Engineering — one of the Nature family's leading journals for translational biomedical research, with an Altmetric score of 43 for this piece indicating broad cross-disciplinary attention — argues that the moment of truly useful large language models has arrived. The editorial states that, barring the use of advanced watermarking strategies, it is no longer possible to accurately distinguish text written by a human from text generated by a large neural network — a threshold the editors note that most machine learning experts and linguists would not have believed reachable only a few years prior. Writing from a biomedical science context, the editorial anticipates that the consequences of LLM proliferation will be far-reaching across research, clinical documentation, and scientific publishing. The piece is notable as an early instance of a major peer-reviewed journal's editorial board officially registering that the distinction between human and machine writing had effectively collapsed in practice. Critics and researchers in computational linguistics, however, caution that passing indistinguishability tests does not imply genuine language understanding, and that LLM outputs in scientific domains remain prone to hallucination, fabricated citations, and subtle factual errors that human reviewers may fail to catch precisely because the prose is fluent.
2023 (March 14)	Business and personal productivity writing	Product integration	Google announces a major step in integrating generative AI into Google Workspace. Starting with Docs and Gmail, trusted testers would gain access to new AI-powered writing tools that help users draft, rewrite, summarize, and adjust tone effortlessly. Over time, similar enhancements will roll out to Slides, Sheets, Meet, and Chat, enabling features like auto-generated visuals, data analysis, and note-taking. Google emphasizes user control, data privacy, and responsible AI design in alignment with its AI Principles. The initiative aims to transform Workspace into a collaborative AI partner, enhancing productivity while preserving human creativity and decision-making.
2023 (March 16)	Workplace communication and document generation	Product integration	Microsoft 365 Copilot is introduced as an AI-powered productivity tool that integrates large language models like GPT-4 with Microsoft 365 apps and user data through the Microsoft Graph. It aims to transform how people work by enhancing creativity, boosting productivity, and upgrading skills using natural language prompts. Embedded in Word, Excel, PowerPoint, Outlook, Teams, and more, Copilot automates tasks, summarizes content, and generates creative outputs. It also introduces Business Chat, which pulls contextual data across apps. Copilot prioritizes security and privacy, using a permission-based model and preserving enterprise compliance.
2023 (March 27)	Educational writing and academic instruction	Commentary	Three NC State University English professors offer diverging assessments of generative AI's implications for writing and education. Chris Anson, Distinguished University Professor of English and longtime director of NC State's Campus Writing and Speaking Program and former chair of the Conference on College Composition and Communication, warns that delegating complex writing tasks to AI bypasses the cognitive processes essential to intellectual development, potentially stunting writers' growth. Huiling Ding, Professor of Technical Communication and a researcher in responsible AI and workforce development, highlights ethical concerns including authorship integrity, plagiarism, and the displacement of professional writers and artists by automated systems. Paul Fyfe, Professor of English and director of NC State's Graduate Certificate in Digital Humanities and a co-investigator on the university's Center for AI in Society and Ethics, takes a more exploratory stance, encouraging structured classroom experimentation with generative AI tools as a way to develop critical AI literacy rather than avoiding them outright. The three perspectives — caution, ethical critique, and critical engagement — reflect the spectrum of positions emerging in writing instruction at the moment of ChatGPT's mass adoption.
2023 (July)	Academic and educational writing	Commentary	The UNC handout Generative AI in Academic Writing offers guidance on using tools like ChatGPT and Microsoft Copilot in academic contexts. It explains how generative AI works by predicting text based on large language models trained on internet data. Potential uses include brainstorming, outlining, summarizing, editing, translating, and drafting transactional communications. The handout warns against over-reliance, highlighting risks like factual inaccuracies, fabricated citations, biased outputs, and violations of academic integrity.
2023 (August 31)	General-purpose writing and content creation	Product launch	Baidu, China's dominant search engine and AI company, publicly launches ERNIE Bot (文心一言, Wenxin Yiyan) to all users following an invited-preview period that began March 16. Developed on Baidu's ERNIE (Enhanced Representation through Knowledge Integration) large language model trained on trillions of Chinese-language web pages, search data, and a knowledge graph of 550 billion facts, ERNIE Bot is explicitly designed around five generative use cases: literary creation, business writing, mathematical reasoning, Chinese language understanding, and multi-modal generation. Over 30,000 enterprise users apply for API access within two hours of the March preview announcement. ERNIE Bot's public launch coincides with the implementation of China's Generative AI Measures, the world's first comprehensive regulatory framework for generative AI products, which requires algorithmic filing and content compliance before public deployment. The launch marks the arrival of large-scale AI writing assistance in Chinese — the world's most-spoken language by native speakers — demonstrating that the generative writing tool phenomenon emerging from English-language markets is simultaneously developing independently in non-Anglophone contexts, with distinct regulatory, cultural, and linguistic constraints shaping its trajectory.
2023 (September 12)	Creative writing	Legal	Michael Chabon, David Henry Hwang, Matthew Klam, Rachel Louise Snyder, and Ayelet Waldman file a class-action complaint against Meta Platforms, Inc., alleging copyright infringement related to the training of Meta's LLaMA large language models. The plaintiffs—prominent authors and playwrights—claim their copyrighted works were copied without consent through inclusion in the Books3 dataset, derived from the shadow library Bibliotik and used as part of LLaMA's training corpus. They argue Meta relied on large quantities of copyrighted books despite restrictive licenses, violating the Copyright Act and the DMCA. The suit seeks relief for unauthorized reproduction and distribution.
2023 (September 24)	Creative writing	Commentary	Scottish outdoor writer and editor Alex Roddie — editor of Sidetracked magazine and a regular contributor to The Great Outdoors — argues in a personal essay that generative AI threatens the essence of creative writing by automating the thinking and emotional processes essential to genuine expression. Writing from the perspective of a working professional writer with a direct livelihood stake in the question, Roddie contends that AI can mimic language but lacks lived experience, intentionality, and individuality — the core of meaningful writing. He warns that outsourcing creativity to AI weakens critical thinking and erodes the human voice, and sees generative AI as fostering conformity and diluting originality. Other writers and researchers, however, argue the opposite: that AI functions as a generative tool that can expand creative possibilities, reduce friction in early drafting, and help writers develop ideas they would not have reached alone — with the quality of the final work remaining the author's responsibility.
2023–2025	General-purpose writing and content creation	Legal	In the United States, federal legislative efforts to require disclosure of AI-generated text produce no enacted law during the first wave of generative AI adoption. The AI Labeling Act of 2023 (S.2691), introduced July 27, 2023 by Senator Brian Schatz (D-HI) with Senator John Kennedy (R-LA) as cosponsor, would require clear and conspicuous disclosures on AI-generated text content — including permanent metadata identification of the generating tool and date — and defines AI-generated content as material "substantially created or modified" such that it "materially alters the meaning or significance that a reasonable person would take away," explicitly excluding minor AI assistance from the disclosure requirement. Enforcement would fall to the Federal Trade Commission as an unfair or deceptive trade practice. The bill is referred to the Commerce, Science, and Transportation Committee and advances no further, expiring when the 118th Congress concludes in January 2025. The bipartisan REAL Act, introduced December 10, 2025 by Representatives Bill Foster (D-IL) and Pete Sessions (R-TX), requires federal agencies to label AI-generated content in official channels but is narrowly scoped to government communications rather than publishing broadly. California's SB 942, signed September 28, 2024 and effective January 1, 2026, becomes the most significant enacted US legislation on AI content labeling, but operates at state rather than federal level. The absence of enacted federal legislation covering AI disclosure in professional writing and publishing — despite repeated bipartisan congressional interest — leaves US writers, publishers, and readers without a consistent legal standard, in direct contrast to the binding EU framework taking effect in August 2026.
2024–2025	Academic and scientific writing	Commentary	Major academic publishers respond to documented AI use in peer review with formal policy frameworks, producing significant variation across institutions. A cross-disciplinary analysis of policies from 439 high-impact-factor and 363 middle-impact-factor journals across 21 disciplines — conducted by researchers Zhongshi Wang and Mengyue Gong and published in Learned Publishing in December 2025 — finds that by August 2025, 83% of high-IF journals and 75% of middle-IF journals have adopted AI policies for peer review, with policies evolving rapidly: 24.5% of high-IF journals revise their policies in just the five months between March and August 2025. The analysis identifies five distinct policy types anchored to major publishers. Elsevier takes the most restrictive stance, banning AI from all peer review and editorial stages including auxiliary tasks such as language improvement. Springer Nature prohibits uploading manuscripts to generative AI tools but does not explicitly ban AI use, requiring only disclosure. Wiley permits limited use, allowing grammar and readability polishing in review reports. The Association for Computing Machinery (ACM) is the most permissive, allowing generative AI to improve review quality provided confidential information is removed. Despite these differences, all five policy types share one consensus: AI cannot replace human judgment on a manuscript's innovation or professional standing. Disciplinary variation is pronounced — science, technology, and medicine disciplines enforce stricter restrictions, while humanities and social sciences adopt more permissive approaches. The Committee on Publication Ethics (COPE) prohibits AI from being listed as an author or fulfilling peer review responsibility, and requires explicit disclosure of AI tool use, version, and purpose.
2024 (February 5)	Literary and creative writing	Commentary	Japanese novelist Rie Kudan, 33, discloses at the acceptance ceremony for the 170th Akutagawa Prize — Japan's most prestigious literary award for emerging fiction writers, established in 1935 — that approximately 5% of her winning novel Tokyo-to Dojo-to (Sympathy Tower Tokyo) consists of text generated verbatim by ChatGPT. The prize committee had praised the novel as "almost flawless" before the disclosure was made. Kudan explains she turned to ChatGPT to find "soft and fuzzy words" that captured the novel's muddled themes of justice and AI, and describes using the tool as a personal confidant for thoughts she could not share with other people. The revelation prompts literary award bodies in Japan to reconsider their submission guidelines. Defenders of Kudan's approach note that the AI-generated passages are largely confined to dialogue attributed to an AI character within the novel, making the use structurally intentional rather than a straightforward substitution of human prose — a distinction that complicates simple narratives about AI replacing authors. The event is among the first high-profile cases of AI-assisted writing winning a major literary prize, and coincides with a similar case in China, where journalism professor Shen Yang used AI to generate a science fiction novel in three hours, winning a national competition without judges being aware of its AI origins.
2024 (March 18)	Educational and creative writing / Professional writing	Research publication	A study by Zhuoyan Li, Chen Liang, Jing Peng, and Ming Yin, presented at the CHI Conference on Human Factors in Computing Systems, explores how people value and experience generative AI-powered writing assistance. In a randomized experiment with 379 participants writing essays or stories, researchers test independent writing, AI editing help, and AI-driven drafting. Participants show they are willing to forgo pay for AI support, especially for creative tasks and full content generation. AI is found to boost productivity, confidence, and grammar but also reduces accountability, satisfaction, and writing diversity, revealing both benefits and risks for education and professional use.
2024 (August 1)	General-purpose writing and content creation	Legal	The European Union's Artificial Intelligence Act (Regulation EU 2024/1689) — the world's first comprehensive AI law — enters into force, with its transparency provisions establishing the first binding legal framework directly affecting AI-generated written text. Article 50 of the Act requires providers of generative AI systems to ensure their outputs are marked in machine-readable formats, and requires deployers using generative AI for professional purposes to clearly label AI-generated text published on matters of public interest — with an exemption for content that has undergone genuine human editorial review and where a natural or legal person bears editorial responsibility. The labeling requirement follows a phased timeline, becoming fully applicable from August 2, 2026. In preparation, the European Commission launches work on a voluntary Code of Practice for marking and labeling AI-generated content on November 5, 2025, publishing a first draft on December 17, 2025, with finalization expected by June 2026. The Code adopts a multilayered approach to machine-readable watermarking covering synthetic audio, images, video, and text. The AI Act's editorial exemption — requiring demonstrable human review rather than nominal oversight — directly shapes how professional writing operations in EU member states structure their AI-assisted workflows.
2024 (October 1)	Scholarly and research writing	Commentary	An article by Charlotte Huff explores how AI tools can support psychology researchers and students with tasks like grammar correction, citation formatting, and idea generation. However, the article warns against over-reliance on AI, emphasizing the need for human authorship, ethical oversight, and transparent citation. APA Style guidelines stress that only humans can be listed as authors and that all AI usage must be properly documented. Risks include misinformation, fabricated citations, detection failures, and bias—particularly against non-native English speakers. Ultimately, Huff points out that AI should augment, not replace, the scholar’s role in the research process.
2024	Academic and scientific writing	Research publication	Quantitative studies begin documenting the scale of AI involvement in academic peer review for the first time. A study of 50,000 peer reviews for computer science conference papers submitted in 2023 and 2024, published in Nature, estimates that up to 17% of sentences are likely written by a large language model. A separate study of peer reviews submitted for the 2024 International Conference on Learning Representations (ICLR) — one of the most prestigious machine learning venues — finds that at least 15.8% of reviews are at least partially AI-generated, with AI-assisted reviews correlating with higher paper scores and acceptance rates, raising concerns about fairness and integrity in the scholarly review process. A 2026 Nature survey of 1,600 academics subsequently finds that more than 50% have used AI tools while peer reviewing manuscripts, confirming that undisclosed AI use in peer review has become a majority practice in at least some academic communities. Researchers also document cases of authors embedding hidden prompts — in white text or microscopic fonts — within manuscripts to manipulate AI-powered review systems into producing favorable assessments, exposing a fundamental vulnerability in automated scholarly judgment.
2025 (January)	Interactive storytelling and fiction writing	Research publication	An article presents a novel interactive system, ChatGeppetto, for AI-assisted narrative generation based on semiotic reconstruction. Drawing on syntagmatic, paradigmatic, meronymic, and antithetic relations, the system enables users to remix and create stories from existing narratives. The AI co-author, powered by ChatGPT and Stable Diffusion, uses abductive reasoning to generate text and visual scenes. A user-friendly prototype supports casual users and writers, aiming to enhance creativity and story coherence. While initial user studies show no statistically significant differences in satisfaction, the semiotic model is praised for enriching emotional and memorable content. Ethical concerns around authorship, bias, and AI’s creative role remain central to future research.
2025 (February 13)	Creative writing	Research publication	A study explores how 18 creative writers intentionally integrate AI—specifically large language models—into their writing practice without compromising core values like authenticity and craftsmanship. Unlike other forms of writing, creative writing is seen as highly personal and choice-driven. While some see AI as reducing creativity to mere prompting, the writers studied develop nuanced, evolving relationships with AI tools. They make deliberate choices about when and how to use AI, building custom workflows that preserve their creative control. These writers see AI not just as a tool, but as a dynamic collaborator, offering insight into ethical and sustainable ways to support creativity with technology.
2025 (February 20)	Professional and creative writing	Commentary	British novelist and science writer Dr Sanjida O'Connell — author of eight novels and four non-fiction books, a Betty Trask Award winner, and a Fellow of the Royal Literary Fund, the UK's principal charitable body supporting professional writers — outlines ten ways writers can use AI to enhance their craft without sacrificing creativity, writing for the RLF's own publication. O'Connell's position carries particular weight as that of a working author who describes initial skepticism rooted in the unauthorized use of her work in AI training data, and who arrived at a pragmatic accommodation rather than outright rejection. Her practical suggestions include using AI for brainstorming, generating writing prompts, locating literary agents, crafting pitches, summarizing research, proofreading, and producing publicity material. She emphasizes prompt engineering as essential and urges writers to retain editorial control and preserve their individual voice, noting that while AI output tends toward the generic, it can still generate ideas that spark original writing. Critics of this accommodationist stance argue that framing AI as merely a useful tool for established professional writers obscures the structural harm it poses to newer or less-established writers — precisely those with less leverage to set the terms of their engagement with AI platforms.
2025 (May 23)	Literary writing and personal expression	Commentary	Australian poet and artist Luke Beesley — author of multiple collections published by Giramondo Publishing, including In the Photograph, shortlisted for the 2024 Prime Minister's Literary Awards — argues in an essay published in The Guardian that writing by hand, specifically with pencil, has become a quiet act of resistance against AI's growing role in literature. Writing from the perspective of a working poet whose own creative process is grounded in handwritten first drafts, Beesley contends that as AI-generated content becomes harder to distinguish from human work, authenticity will increasingly reside in process and material evidence — the notebook, the pencil stub, the physical archive of creation — rather than in the finished text alone. He suggests these traces may soon function as proof of genuine human authorship amid widening concerns about trust and plagiarism in literary culture. Others challenge this framing, arguing that fetishizing handwriting as a mark of authenticity is aesthetically conservative and practically exclusive — privileging a mode of composition inaccessible to many writers due to disability, circumstance, or working habit — and that the integrity of literary work should be judged by its qualities rather than its production method.
2025 (May 26)	Investigative and professional journalism	Research publication	A study suggests journalism is not dying but evolving with AI. Despite fears of newsroom collapse, Danish labour market data show little evidence of job losses linked to large language models. Public anxiety remains high: surveys reveal most Americans expect declining news quality, job cuts, and misinformation. Yet investigative work demonstrates AI’s constructive role. Outlets like The Wall Street Journal, The Washington Post, and Associated Press use machine learning and geospatial tools to uncover stories, proving AI augments reporting with speed and depth while preserving human judgment.
2025 (June 12)	Long-form story generation	Research publication	A paper introduces SCORE, a framework designed to enhance coherence and emotional depth in AI-generated stories from Large Language Models (LLMs). SCORE identifies narrative inconsistencies by tracking key items, generating episode summaries, and using a Retrieval-Augmented Generation (RAG) approach with TF-IDF and cosine similarity. It addresses common LLM issues such as inconsistent character behavior and emotional tone. Inspired by memory mechanisms in generative agents, SCORE evaluates character consistency, emotional flow, and plot logic. Tests on LLM-generated narratives show that SCORE significantly improves coherence and stability compared to standard GPT models, offering a more structured method for refining long-form AI storytelling.
2025 (June 27)	Book publishing and literary labor	Commentary	Over 70 authors publish an open letter on Literary Hub urging major U.S. publishers to limit their use of generative AI, with the initiative quickly gathering over 1,100 signatures. The letter's prominent signatories include Dennis Lehane — Boston-born crime novelist and screenwriter whose works including Mystic River, Gone Baby Gone, and Shutter Island have been adapted into major films — and Lauren Groff — three-time National Book Award finalist and New York Times bestselling novelist whose Fates and Furies was named President Barack Obama's favorite book of 2015. The letter demands commitments from publishers not to release AI-generated books, replace human workers with AI, or train AI on copyrighted material without consent, and calls for the continued use of human narrators and translators. Organizers frame the letter as a response to an existential threat to authors' livelihoods, particularly following recent court rulings favoring AI companies in copyright disputes. The initiative is notable for its speed of momentum and for targeting publishers rather than AI companies directly — holding the industry intermediary accountable rather than the technology developer. Of the major publishers approached, only Simon & Schuster responds publicly, affirming its commitment to protecting authors' rights; the silence of the other major houses is itself interpreted by organizers as meaningful. Publishers and AI proponents counter that human-AI collaboration in book production is not categorically different from other forms of editorial and technological assistance authors have long accepted, and that blanket prohibitions risk constraining tools that could benefit writers as much as harm them.
2025 (June 30)	College and academic writing	Commentary	Hua Hsu — staff writer at The New Yorker, professor of English at Bard College, and 2023 Pulitzer Prize winner for his memoir Stay True — explores how generative AI is reshaping college writing and, by extension, higher education itself. Drawing on classroom observations and interviews with students and educators, Hsu finds that students increasingly rely on tools like ChatGPT for everything from formal essays to personal communication, often bypassing the writing process rather than using AI to augment it. Some educators respond with in-class exams and handwriting exercises; others embrace AI as a legitimate learning aid. Hsu frames students' behavior not primarily as cheating but as a rational response to structural conditions — an efficiency-oriented, consumer-minded educational culture that has failed to articulate clear goals for what writing is supposed to develop in the first place. Writing from the dual vantage of a literary journalist and a practicing college teacher, his account carries particular weight as an insider perspective on the crisis. Those who take a more optimistic view argue that AI's pressure on traditional writing assignments could prompt long-overdue reform of rote assessment practices, and that learning to collaborate effectively with AI tools is itself a transferable skill worth cultivating.
2025 (August 28)	Personal messaging	Product launch	WhatsApp introduces “Writing Help”, an AI-powered assistant that offers in-app suggestions to refine your messages’ tone, style, and clarity. Users simply compose a message, tap the pencil icon, and choose from options like professional, funny, rephrase, supportive, or proofread to enhance their text before sending. Powered by Meta’s Private Processing technology, all suggestions remain encrypted and private—neither WhatsApp nor Meta can access the original message or AI output. The feature is first being rolled out in English to select regions, with broader availability expected later in the year.
2025 (September 10)	Creative writing	Commentary	Rice University launches a course that examines how generative AI can both inspire and challenge creative writing. Rather than outsourcing storytelling to ChatGPT, students explore ways to incorporate or resist AI’s influence, using it to spark ideas while confronting its limitations. The class engages with critical essays and AI-generated texts, while also addressing ethical issues such as copyright and environmental costs. Associate teaching professor Ian Schimmel emphasizes that grappling with AI’s flaws and controversies fosters deeper reflection on creativity and technology.
2025 (October 29)	Literary and creative writing	Legal	George R.R. Martin sues OpenAI and Microsoft as part of a broader class-action case alleging unauthorized use of copyrighted books to train AI models. A federal judge allows the lawsuit to proceed after ruling that a ChatGPT-generated outline for a fictional Game of Thrones sequel, A Dance With Shadows, could be substantially similar to Martin's work. The output includes elements resembling A Song of Ice and Fire, prompting further legal examination.

Meta information on the timeline

How the timeline was built

The initial version of the timeline was written by Sebastian.

Check [1].

Funding information for this timeline is available.

Feedback and comments

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

• FIXME