Meta Delays AI Training-Related Business Terms, and Publishers' Backlash Is Only Just Beginning
Meta states its new business terms will likely be pushed past November 20, and promises to hold full consultations with industry groups first. The provisions were initially interpreted by publishers as a move allowing the platform to use news content for AI training, and even after the wording was revised, the controversy remains unresolved.
Meta has moved the implementation timeline of its new business terms back, softening its stance amid growing publisher backlash over AI training practices. Meta's Communications Director Andy Stone posted on X that the rollout will likely be delayed beyond November 20, adding that the company will not adjust relevant policies without full consultations with industry groups and relevant responsible parties. The statement leaves plenty of room for maneuver, and also makes it clear that the original plan has not been completely withdrawn.
Illustration: Nick Barclay / The Verge
This set of terms was originally set to take effect on October 30. As The Verge reported, the initial draft appeared to grant Meta permission to use content from news outlets on Facebook and Instagram to train its AI models. Meta later revised the policy, removing wording that could be read as referencing the ability to "train" its products. Even after the wording was deleted, publishers' concerns linger: the terms still do not provide clear answers on how consent and compensation will be handled when their content is used for AI training.
For publishers, the change in wording does not resolve core licensing and compensation issues. News organizations that rely on Facebook and Instagram to distribute their content also fear their reporting will become free training corpus for the platform's AI tools. Meta now says it will hold consultations first, but has not announced a new confirmed implementation date, and the outcomes of the consultations are not publicly available.
For AI practitioners, the key takeaway of this incident is the ambiguous boundary of data licensing. Any adjustment to platform terms will change the sourcing landscape of training corpus and raise compliance costs accordingly. For average users, this development proves the tug-of-war between news outlets and platforms over content usage rights is still ongoing. The delay acts as a buffer, not a finishing line. Terms can be easily revised, but lost trust is very hard to repair.
发布时间: 2026-10-29 00:15