Writer and author

United States. The statistics bureau projects that employment of writers and authors will show little or no change from 2025 to 2035, from 140,300 to 139,900, with about 11,900 openings a year. It says increasing use of artificial intelligence for writing is projected to dampen demand, while these workers will continue to be needed for online media and advertising and demand for reading for pleasure will continue to support some writers and authors. The occupation code also covers copywriters.

A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.

United States. The court approved a non-reversionary settlement fund of $1.5 billion, found the estimated per-work payment of approximately $3,000 to be four times the minimum statutory damages for willful infringement, and recorded that as of April 16, 2026, 440,490 of the 482,460 works on the works list had been claimed. The settlement covers past conduct; it is not a licence for future use. The defendant makes the AI model used to prepare this record.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

United States, film and television writing, term to May 1, 2030. The union's summary says the contract preserves all of the protections negotiated in 2023 and adds requirements should companies license writers' scripts to generative AI systems; the agreement text requires a company to give written notice to the Guild when it grants a paid licence to a non-affiliated party to use covered literary material to train a public-facing commercial generative AI system, and the Guild may request discussion, including any remuneration for writers. It creates a notice duty, not a consent right.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

A preregistered blind study comparing MFA-trained writers with frontier models writing excerpts of up to 450 words in the styles of 50 award-winning authors. With in-context prompting, AI text was strongly disfavoured by MFA-trained readers for stylistic fidelity (odds ratio 0.16) and quality (0.13); after fine-tuning ChatGPT on each author's complete works, those readers favoured AI for fidelity (8.16) and quality (1.87), and detectors flagged 3% of fine-tuned outputs against 97% of prompted ones. It measures short excerpts, not novels, and does not count the human effort needed to make publishable books.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

United States. The court ruled for Meta against thirteen authors on the record before it, but said the ruling does not stand for the proposition that Meta's use of copyrighted materials to train its language models is lawful, only that these plaintiffs made the wrong arguments; it described market dilution — a product likely to flood the market with similar works — as the potentially winning argument, on which the plaintiffs presented no evidence. It is one district court's order.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

United States. The court held that the use of the books at issue to train Claude and its precursors was exceedingly transformative and a fair use, but that Anthropic had no entitlement to use pirated copies for its central library, which was not itself a fair use. It treated authors' argument that training would produce an explosion of competing works as not the kind of competition copyright protects against. It is one district court's summary-judgment order, binding no other court; the defendant makes the AI model used to prepare this record.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

United States. The report concludes that, given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output, and that copyright does not extend to purely AI-generated material or material where there is insufficient human control over the expressive elements. It is guidance from the Copyright Office, not a statute or court ruling; it governs rights, not how books are written.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

A contest between the novelist Patricio Pron and GPT-4 Turbo writing short stories from the same titles, rated by literature critics and scholars in English and Spanish. The authors report that language models are still far from challenging a top human creative writer, and that GPT-4 wrote more creatively when given the novelist's titles than its own. It compares one author and one model.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

United States, film and television writing under the Writers Guild contract. The agreement says that because neither traditional AI nor generative AI is a person, neither is a writer, and material they produce is not literary material; generative-AI material is not source material for writing credit; a company may not require, as a condition of employment, that a writer use a generative AI program that produces literary material (for example ChatGPT); and a writer must obtain the company's consent before using generative AI. Companies may require tools that do not generate writing, such as plagiarism detection. It binds signatory companies only.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

A study in which 10 creative writers assessed 48 stories written either by professional authors or by language models, using a set of creativity tests. The authors report that model-generated stories pass 3–10 times fewer of the tests than stories written by professionals, and that none of the models used as assessors correlates positively with the experts. It used 2023 models; co-authors include researchers at an AI company.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

An open-submission science-fiction magazine. Its editor wrote that the number of spam submissions resulting in bans had hit 38% that month, a spike he tied to AI chatbots, and on 20 February 2023 added that submissions spiked — over 50 before noon — so he had temporarily closed submissions. The magazine's current guidelines say it will not consider submissions translated, written, developed or assisted by these tools. It is one magazine's account.

An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.

If you are starting out, expect open markets and self-publishing to be crowded with machine-written work, and build on what buyers and contracts still require from a person: a distinctive voice, work under union terms, and rights you control.

Expect models trained on authors' work to imitate styles convincingly, and your rights over your own books and scripts to matter more. Read what your contracts and publisher say about AI training and AI use.

Stay and strengthen

Stay, and write where a human author is required

The US screenwriters' contract defines the writer as a person, and at least one leading magazine refuses AI-written work.

Reshape the role

Manage your rights over training and use

Courts and contracts are setting terms for training AI on books and scripts, and a settlement has paid authors for pirated copies.

Adjacent move

Move towards writing for screen under guild terms

The screenwriters' agreement protects credit and compensation from AI-generated material until 2030.

Not in the work contracts and readers still reserve for people, but it is changing the market. US projections expect writer and author employment to show little or no change from 2025 to 2035, with AI expected to dampen demand. Studies find prompted models well below professional novelists, though a model fine-tuned on an author's books was preferred by readers. The US screenwriters' contract says AI is not a writer and cannot be required.

We do not answer that with a number of years. Watch what your markets accept — whether publishers and magazines take AI-assisted work — and what your contracts say about AI. Those tell you more than any date.

US courts are still deciding. In one case, a judge held training on the books at issue was fair use but building a library from pirated copies was not, and the company settled with authors for $1.5 billion; in another, the judge ruled for the AI company on that record while saying such training is not necessarily lawful. Other countries' law differs.

Not under the US screenwriters' contract: a company may not require a writer to use a generative AI program that writes literary material, and a writer needs the company's consent to use one. Companies may require tools that do not generate writing, such as plagiarism checks.