Meta Signals Another Delay to Controversial Business Terms as Publisher Backlash Grows

Meta is preparing to delay its new business terms for a second time, after publishers and lawmakers pushed back hard over fears the policy would let the company train AI models on news content shared through Facebook and Instagram.

Meta communications head Andy Stone confirmed the likely delay in a post on X, saying the company does "not plan to make policy changes without a full consultation with industry groups and leaders."

The saga started with terms that were originally set to take effect October 30th. As initially written, they appeared to give Meta the green light to train its AI systems on content from news organizations using its platforms. That set off alarm bells across the publishing industry, which has spent the past few years watching tech companies hoover up its content to build AI products, often without compensation or consent.

Meta tried to defuse the situation by rewriting the terms to strip out language explicitly mentioning "training" on scraped content. But according to Nieman Lab, legal experts who reviewed the revised language said it still didn't fully rule out that Meta could use publisher content for AI training. The ambiguity wasn't enough to satisfy critics, and Meta pushed the effective date to November 20th.

Now even that deadline looks shaky. In his statement, Stone said Meta expects the consultation process "could extend beyond November 20," in which case the company would delay implementation again. Until a new policy is finalized, Meta's existing terms stay in place.

The pressure campaign hasn't come from just one direction. The News/Media Alliance, a trade group representing publishers, has been vocal in its opposition. And the pushback has crossed party lines in Washington. Sen. Marsha Blackburn, a Tennessee Republican, accused Meta of "attempting to exploit American creators and businesses" with the updated terms. Sen. Adam Schiff, a California Democrat, has called on Meta to reverse the change entirely.

That kind of bipartisan alignment is notable. Tech regulation fights usually split along predictable lines, with Democrats focused on issues like antitrust and content moderation while Republicans focus on censorship claims. AI's use of copyrighted journalism, though, has managed to unite lawmakers who agree on little else, in part because it touches both economic fairness and the survival of local and national news outlets already squeezed by declining ad revenue and years of traffic lost to tech platforms.

For Meta, the repeated delays suggest the company either misjudged how fiercely publishers would react, or is genuinely trying to find language precise enough to hold up to legal scrutiny while still preserving flexibility for its AI ambitions. Either way, the back-and-forth highlights a tension playing out across the AI industry: companies need vast amounts of text to train large language models, and news archives are some of the richest, most current sources available. Publishers, meanwhile, are increasingly unwilling to hand that material over for free, especially after watching search and social traffic already erode their revenue.

Meta hasn't said what a revised version of the terms might look like, or how long the "full consultation" with industry groups might take. For now, the practical effect is that nothing changes. Meta's current business terms remain in force past the original deadline, and publishers get more time to make their case before any new rules take hold.

What happens next likely hinges on whether Meta can find terms specific enough to satisfy legal experts and publishers that AI training isn't on the table, without undercutting whatever flexibility the company wants for its broader AI strategy. Given how this has gone so far, don't expect a quick resolution.