Meta is preparing a sweeping update to its commercial terms of service that will take effect on October 30, requiring business accounts to grant the company permission to train its AI models on content from their linked websites, according to a report from Nieman Lab. The change extends beyond social media posts to any content on sites that businesses link to from their Meta accounts.
Legal experts have raised alarms that the update, which cannot be opted out of, effectively gives Meta a broad license to crawl and use website content for AI training. Businesses that do not agree are advised to stop using Meta’s services. The development is particularly significant for news publishers, many of whom are already engaged in legal battles with other AI companies, such as OpenAI, over whether such training without licensing constitutes fair use. Meta’s new terms could preempt those disputes by granting itself contractual permission, regardless of how courts rule on the fair use question.
The change comes amid ongoing uncertainty over the legality of AI training on copyrighted material, with publishers seeking to negotiate licensing deals. By tying AI training rights to its commercial terms, Meta may sidestep the need to pay for access to large volumes of web content that it previously could have been required to license.