Meta Delays AI Terms, but Publishers Say the Scraping Threat Remains
Meta Delays AI Terms, but Publishers Say the Scraping Threat Remains
Meta’s dispute with publishers began with business terms that were due to take effect on October 30. The original wording said Meta could use content from websites linked to business accounts to “develop, research” and “train” its products — language that alarmed news organizations already fighting to protect their work from AI harvesting.
After a fierce industry response, Meta removed the explicit reference to training. But critics argued that the rewrite changed the optics more than the substance. Cornell digital-law professor James Grimmelmann said the remaining provision allowing use of material “associated with your products or services” could still let Meta scrape “entire sites.” Media and intellectual-property lawyer Matt Topic made the same broader point: permission to “develop” an undefined “business product” could readily encompass AI training.
The political backlash widened as well. Sen. Marsha Blackburn, the Tennessee Republican, said the updated terms suggested Meta was seeking “sweeping authority to scrape content from any website associated with Facebook and Instagram business accounts.”
Publishers’ trade group the News/Media Alliance wrote to CEO Mark Zuckerberg demanding revisions, while advertisers and publishers in the EU and UK pressed regulators to intervene. Meta then pushed enforcement back to November 20.
Meta spokesperson Daniel Roberts said the delay would give the company more time to explain what had — and had not — changed. The company’s defense rests on robots.txt, the long-standing web file through which site owners can tell crawlers their preferences. Roberts said the update “is not about scraping news sites” and does not create rights that override those instructions.
That assurance has not settled the argument. Meta frames the terms as a clarification of existing web-crawling practices; publishers and their legal allies see ambiguous language that could still turn connected business websites into a vast reservoir for AI development.
Write a comment