Meta, led by Mark Zuckerberg, is implementing a broad new policy for dealing with manipulated media and AI-generated images, videos and audio on Facebook, Instagram and Threads. The move follows public criticism and recommendations from the company's Oversight Board, with the aim of giving users greater transparency without unnecessarily restricting freedom of expression.
Rather than broadly removing edited or AI-generated content that does not otherwise violate its Community Standards, such as rules against harassment or election interference, Meta has opted for a less restrictive approach based on informational labels and added context.

The company's original policy, adopted in 2020, focused mainly on videos manipulated to make people appear to say things they had never said. But the rapid development of increasingly realistic AI-generated images and audio prompted Meta to overhaul its approach.
Key changes to Meta's labeling policy
The company has begun applying labels indicating the use of artificial intelligence to a wider range of images, videos and audio.
The content was initially labeled "Made with AI," but Meta changed the wording to "AI info" after users complained that the label was also appearing on content that had undergone only minor AI-assisted edits, such as retouching. The revised wording is intended to provide a more accurate indication of the extent to which AI was used.


The labels are based on industry-standard signals embedded in files created or modified with AI tools, as well as cases in which users themselves disclose that they are uploading AI-generated content.
False or misleading content reviewed by independent fact-checkers may receive an additional layer of intervention, including being labeled and shown lower in users' feeds so fewer people see it.
Content that does not violate Meta's rules, however, generally remains on the platforms with relevant labels and context, allowing users to assess it for themselves.



