Meta drops AI training clause after news publisher backlash
Meta has deleted wording that critics said let it train AI on scraped news content. Here is why the retreat matters for Indian publishers and the ANI case.
The News
Meta has removed a clause from its commercial terms that critics read as giving the company licence to use content from third-party websites, news publishers included, to train its AI systems. It is the second revision to the same terms within days, following a wave of criticism.
The deleted passage said Meta "may use any content from such domains, web pages, or URLs" to develop, research and train its products and services offered under the commercial terms. Neiman Lab and James Grimmelmann, a professor of digital and information law at Cornell, argued the language could let Meta train on scraped material for free and override the instructions publishers place in their robots.txt files.
Meta initially denied any shift. Andy Stone, its vice president of communications, and spokesperson Daniel Roberts said "There is no change to how Meta uses web crawlers." In a follow-up statement, Roberts said the consolidated crawling terms applied only to Meta's ads and business products, that robots.txt remains the way for site owners to state their preferences, and that the update grants Meta no new rights over those preferences.
Why It Matters
The episode shows how quickly the fine print of AI training has become a reputational flashpoint. Large platforms are hungry for fresh, well-edited text, and news archives are among the most valuable sources available. Any wording that appears to widen access without payment now draws immediate scrutiny from lawyers, academics and trade press.
There is a clear precedent. In August 2023, Zoom faced a similar backlash over terms of service that seemed to permit training AI on customer content, and it rewrote them within days. The pattern is repeating: an expansive clause appears, outside experts flag it, and the company retreats while insisting nothing has changed in practice. The retreat matters less than what it reveals, namely that the legal basis for training on the open web is still being negotiated in public, one terms page at a time.
Indian Angle
For Indian publishers, the timing is pointed. News agency ANI has an ongoing copyright suit against OpenAI in the Delhi High Court, filed in November 2024, over the alleged use of its content to train ChatGPT, and several Indian media groups have sought to join the proceedings. Meta's climbdown strengthens the argument that consent and licensing, not silent terms updates, should govern how news is used for training.
Policy is moving in the same direction. A committee set up by the Department for Promotion of Industry and Internal Trade (DPIIT) has been examining how the Copyright Act, 1957 should treat AI training, including licensing models under which developers would pay rights holders. Indian publishers, many of whom depend on Facebook and Instagram for referral traffic, will be watching whether global platforms accept such frameworks or resist them.
There is a practical lesson for Indian businesses too. Startups and enterprises that advertise on Meta or build on its business tools should read revised commercial terms closely, since data-use clauses tucked into ads agreements can reach well beyond advertising. Publishers should also audit their robots.txt rules and consider explicit licensing terms rather than relying on the file alone.
FAQ
What exactly did Meta remove?
Meta deleted wording from its commercial terms stating that it could use content from domains, web pages or URLs to develop, research and train its products and services. Critics said this could let Meta train AI on scraped material without payment. The clause no longer appears on the commercial terms page.
Does this change how Meta crawls the web?
According to Meta, no. Its spokesperson said the consolidated crawling terms only ever applied to ads and business products, that content owners can still use robots.txt to signal their preferences, and that the update does not give Meta rights that supersede those preferences.
What does this mean for Indian news publishers?
It reinforces the case that news content used for AI training should be licensed rather than taken by default. With ANI's case against OpenAI before the Delhi High Court and DPIIT examining AI and copyright, Indian publishers have a stronger footing to demand clear consent and payment terms.
Where can I read the original report?
The Verge published the original report on 17 October 2026, quoting the removed clause and Meta's statements in full. The link is below.
This story was reported by The Verge. Read the full original coverage at The Verge.