Meta plans to tag AI-generated content more broadly

Meta said that the current method of labeling content generated by artificial intelligence is too narrow, and will soon apply the “Made by Artificial Intelligence” badge to a wider range of video, audio and imaging. Starting in May, it will attach the label to media when it detects industry-standard AI image indicators or when a user confirms they are uploading AI-generated content. The company may also apply labels to posts flagged by fact-checkers, although it may lower the ranking of content identified as false or altered.

The company announced the measure after an oversight committee made a decision regarding a video that was maliciously edited to depict President Joe Biden inappropriately touching his granddaughter. The oversight board agreed with Meta’s decision not to remove the video from Facebook because it did not violate the company’s rules on manipulated media. However, the board recommended that Meta should “quickly reconsider this policy given the number of elections in 2024.”

Meta said it agreed with the board’s “recommendation that providing transparency and additional context is now a better way to address the issue of manipulated media and avoid the risk of unnecessarily restricting free speech, so we will keep this content on our on the platform so we can add tags and backgrounds.” The company added that in July it would stop removing content purely for violating its manipulative video policy. “This timeline gives people time to understand the self-disclosure process before we stop removing a small percentage of manipulated media,” Monika Bickert, Meta’s vice president of content policy, wrote in a blog post.

Meta has been labeling the realistic images created by users using Meta AI tools as “Imagine with AI.” Meta said the updated policy goes beyond the oversight committee’s labeling recommendations. “If we determine that a digitally created or altered image, video, or audio poses a particularly high risk of materially deceiving the public about an important issue, we may add a more prominent label so people have more information,” Bickert wrote. and context.”

While the company generally believes transparency and allowing appropriately tagged AI-generated photos, images and audio to remain on its platform is the best way forward, it still removes material that violates the rules. “We will remove content if it violates our policies against voter interference, bullying and harassment, violence and incitement, or any other policy in our Community Standards, whether it was created by artificial intelligence or an individual,” Bickert noted. of.”

The Oversight Board told Engadget in a statement that it was pleased to have Meta on board. It added that it would review the company’s implementation in subsequent transparency reports.

“While it is always important to find ways to protect free speech while preventing clear offline harm, it is especially important to do so in the context of such an important year with elections,” the committee said. “We are therefore pleased that Meta is testing To the AI ​​image metric or when people indicate that they have uploaded AI content, a wider range of video, audio and image content will start to be labeled as ‘Made with AI’. This will provide people with a better experience.” For more types Manipulated Media provides context and transparency while removing posts that otherwise violate Meta’s rules. “

Updated April 5 at 12:55 pm ET: Added Oversight Committee comments.