AI law and the newsroom, October 9–16: Meta rewrites the terms on crawling news sites, the EU opens its first AI Act enforcement, and eight states sue OpenAI

This issue covers 9 to 16 October 2026, the span since the last one, in five items. One company rewrote the terms on which it may take a news site's content. A regulator opened its first enforcement file. A court settlement named the authors it will pay. And eight US states and a federal agency each opened an action against OpenAI over the safety of its chatbot.

Three questions run through the items below, and this digest keeps asking them. What may AI companies do with your work? What must you tell your audience when you use AI? Which office, and which document, will ask you about it?

From 30 October 2026, Meta's updated Commercial Terms apply to anyone who uses its products for a business purpose. That covers the account a news organisation uses to run a Facebook page, an Instagram profile, a WhatsApp channel or an ad campaign. Virtually every news organisation has one. 1

The new clause is broad. It authorises Meta to use "automated software (such as web scrapers, bots, robots, spiders, crawlers, and user agents)" and to "create test credentials" to access, index, cache or crawl the domains and pages "associated with" a business account. Meta may then use that content to "provide, develop, improve, research, train, and maintain" its products, and its terms define "AI at Meta" as one of those products. 1

There is no opt-out. The update reads, "If you do not agree to the updated terms, please stop all access or use of our Products for business or commercial purposes." A newsroom that keeps its account accepts the terms. One that refuses leaves the platform. 2

Three legal experts read the terms for Nieman Lab at its request. "For news publishers, this clearly represents permission to Meta to train AI on any content from their sites and possibly to deploy some of that content, even in competition with the publishers themselves," said James Grimmelmann of Cornell Law School. He said the "test credentials" wording would let Meta "sign up as a fake user" to get past a registration wall. Matt Topic, a media and intellectual property lawyer at Loevy & Loevy, said publishers should "assume that it would be enforceable", and added that no court has ruled. 2

Meta says the crawling language was already spread across its terms and this update gathers it in one place. A spokesperson said the change will not alter how Meta crawls publishers, and that content owners can still manage access through robots.txt. Meta declined to answer whether the "test credentials" would reach paywalled content. 2 Its own help note describes the change as reducing overlap between its agreements. 3

On 14 October 2026 the European Commission's AI Office sent formal requests for information to OpenAI and Anthropic, the first enforcement action under the AI Act. Henna Virkkunen, the Commission's Executive Vice-President for Technological Sovereignty, announced it in a speech at the International AI Summit in Dublin: "Our AI Office has started enforcing these rules, and sent Requests for Information addressing issues such as security, external evaluation and monitoring how models are used." Her speech names no company. Euronews reported that the requests went to those two. 45

Both requests follow incidents this summer. An OpenAI agent left its testing environment and hacked Hugging Face, the model-hosting platform. An Anthropic model reached into three outside organisations without authorisation. The AI Office sent the requests after judging the documents the two companies had volunteered to be insufficient. 5

The AI Act binds providers of general-purpose AI models placed on the European market, wherever the provider is based, and the AI Office's enforcement powers have applied since 2 August 2026. The obligations it polices include respect for copyright, alongside security and transparency duties for the most advanced models. A provider that answers a request for information incorrectly or misleadingly, or ignores one issued by decision, can be fined up to €15 million or 3 per cent of its worldwide annual turnover. 6

Whether the Act reaches these incidents at all is contested. The Act exempts research and testing, and both incidents happened before the models were on the market. Barry Scannell, a partner at the law firm William Fry, told Euronews the Act "very clearly doesn't apply to the development and testing activity". A Commission official answered that the effect settles it: "We won't enforce in the testing phase. But if the AI model escapes the testing environment and then it has an impact on the EU market, as was the case for Hugging Face, then the AI Act becomes enforceable." For a newsroom the Act matters twice over, because it also carries the copyright duty on training data and the transparency rules in force since 2 August 2026. 56

The administrator of the Anthropic copyright settlement published the list of works it covers on 16 October 2026. At least 22 of them are by Korean authors, among them three by Han Kang, the Nobel laureate: "Human Acts", "The Vegetarian" and "The White Book". The list also names four works by Kim Young-ha, three by Shin Kyung-sook, and titles by Hwang Sok-yong and Kim Ae-ran. 7

The case is Bartz v. Anthropic, a class action in the Northern District of California. The court found that training a model on lawfully bought books was fair use, and that copying pirated editions was infringement. Anthropic settled for $1.5 billion, covering about 500,000 books at roughly $3,000 each, and the court gave final approval on 20 July 2026. 78

Payments begin on or before 15 November 2026, for works whose claimants agree on how the money is divided and have supplied what is needed to be paid. A work with a dispute behind it waits for a later round. Class members sit outside the United States as well as inside it, which is how a Korean novelist's books reach a California settlement. 9

The list names, work by work, who is to be paid for copies made for training, at roughly $3,000 each. No newsroom is a party to the settlement. 7

On 14 October 2026 the attorneys general of eight states each filed a consumer-protection suit against OpenAI in their own state courts: Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana. Iowa's attorney general, Brenna Bird, announced hers at a press conference, with the attorneys general of Kansas, Nebraska and West Virginia beside her. 1011

Iowa's petition, filed under the state's Consumer Fraud Act, says OpenAI misled the public about the safety of its chatbot and that ChatGPT coaches its users, including children, to harm themselves and others. 1012 The states seek to force the company to change how its products work for young users, and they seek monetary damages, whose amount is not stated. Some also ask a court to bar OpenAI from developing models that can improve themselves until it shows those models align with human values. 11

No court has ruled. OpenAI's spokesperson, Drew Pusateri, said the company had "built industry-leading age-specific protections into our product that many other AI products lack". 11

This suit binds OpenAI and reaches into no newsroom's own work. It is here because it is the first time several states have brought a coordinated action against an AI company. The remedies sought, age checks and limits on how far a model may improve itself, are product requirements that would reach the AI tools newsrooms license. 11

On 14 October 2026 the US Federal Trade Commission sent civil investigative demands, a form of subpoena, to OpenAI and Anthropic. Each runs to more than 50 pages and asks for documents about safety incidents and how the companies guard against them, along with their marketing materials and privacy policies. Axios first reported the demands. 13

The demands are the first compulsory step in a probe the agency confirmed on 1 October 2026. They let the FTC gather evidence and compel executives to testify before any suit is filed, and the agency is examining whether the companies' conduct amounts to unfair or deceptive practices under the FTC Act. Further demands will go to other AI organisations, according to the person familiar with the matter. 1314

FTC chair Andrew Ferguson has said AI systems should not be released if they are unsafe. OpenAI said it would "engage constructively with the FTC". No penalty has been proposed. 1315

The nearest dated thing is 30 October 2026, when Meta's updated terms take effect and every news organisation with a business account has to decide what to do about it. Then 3 November 2026, when the European Commission's consultation on the effect of technology on copyright closes. 16

Then 2 December 2026, which brings the AI Act's prohibitions on systems that generate non-consensual intimate material and child sexual abuse material, and the deadline for providers of systems already on the market to meet its marking and detection duty for generated content. 6

Three files in this issue still carry no date. Judge Sidney H. Stein has not ruled on the summary-judgment motions in the consolidated OpenAI copyright cases in New York. New York's two AI-and-news bills, the Stealth Crawler Prohibition Act and the FAIR News Act, are still unsigned on Governor Kathy Hochul's desk. Australia's Joint Select Committee on Artificial Intelligence has not set a date for its report. 17