Meta has put on hold an update to its commercial terms that would have required business accounts to let it use content from their own domains, web pages and URLs. A notice on Meta's page says the updates to the Commercial Terms, Product Catalog Terms, Self-Serve Ad Terms and Ad Creative Generative AI Terms will not take effect on 20 November, that existing terms stay in force, and that Meta is still consulting stakeholders and will say more when it can.
“Maintaining a business page or advertising a subscription should not confer general rights over a publisher's website and archive”
Angela Mills Wade, executive director, European Publishers Council (to Reuters), to Reuters · ppc.land ↗
The disputed wording let Meta use that content to provide, develop, improve, research, train and maintain the integrity of its products. Publishers and advertisers read it as an unlimited right to extract their online content for AI training. According to Silicon UK, Meta denied that verbally but offered no written assurance. A Meta spokesperson said content owners could still use robots.txt, though PPC Land notes the terms preview never mentioned it. Under Section 7 of the preview terms, carrying on using Meta's products after notice of an update counted as acceptance, so a business could not accept the terms for its page while refusing them for its website.
The pause is the second step back. On 23 October Meta moved the start date from 30 October to 20 November, saying it needed time to explain the changes, and 404 Media, citing Nieman Lab, reports that it then dropped the word "train" and limited the uses to advertising and business products. On 28 October it said it would consult industry groups, with the consultation expected to run past 20 November.
About Meta
This is not the first time Meta has paused a plan to use people's data for AI. In June 2024 it paused plans to train its AI on the public posts of adult Facebook and Instagram users in the EU and UK after the Irish Data Protection Commission and the UK's Information Commissioner's Office pushed back. The change had been due on 26 June, and the privacy group NOYB had filed 11 GDPR complaints.
A month later, in July 2024, Meta began running a web crawler called Meta External Agent to collect public web data for AI training. It was disclosed only through an update to a developer page, not announced. Meta said it trains its generative AI models on publicly available online content and had updated its guidance for publishers who want to exclude their domains.
7
Models indexed
Models on AIPopDB
Complaints had been building. The Movement for an Open Web wrote to the European Commission and the UK Competition and Markets Authority on 22 October asking for interim measures, and the European Publishers Council, Germany's BDZV and the Association of the Free Press complained to regulators too. BDZV called the terms "unscrupulous theft", and the Publishers Council wanted a suspension rather than a delay. A Norwegian court granted a temporary injunction suspending the terms after a case backed by the Norwegian Media Businesses' Association and 180 media companies, and the Competition and Markets Authority said it would open its own investigation. The US arbitration opt-out window would have closed around 20 December.
Sources · 3