Meta is relaxing its rules against content generated by artificial intelligence. Instead of deleting AI-generated photos, audios and videos, they will be tagged with warnings in the future.
Facebook parent company Meta will allow more photos, audios and videos with warnings generated or manipulated by artificial intelligence (AI) on its platforms instead of deleting them. Meta wants to label content “Made with AI” based, among other things, on the metadata of the files. In addition, in-house technology will be used to automatically recognize posts created with AI.
“We believe that transparency and providing additional information is a better way to handle this type of content,” Meta’s vice president of content policy, Monika Bickert, wrote in a blog post. The selection will be based on “industry-standard AI indicators” or on the basis of user information. Bickert wrote that even clearer labeling could be used for particularly critical content.
Policy violationmessages will continue to be deleted
The company noted that content that violates rules on voter influence, bullying, hate speech, harassment, violence and incitement, or other policies will continue to be deleted – regardless of whether it is human- or AI-made.
Meta’s change follows a recommendation from its independent oversight board, which independently reviews Meta’s content moderation decisions. The “Oversight Board” had warned, among other things, against the publication of misleading, AI-generated content in view of the upcoming presidential election in the USA this year. At the same time, the committee also expressed concerns that deletions could restrict freedom of speech.
Labeling from May, no more deletion from June
The new labels are scheduled to be introduced in May and will apply to Facebook, Instagram and the short message service Threads. In July, Meta plans to abandon its previous more restrictive approach to AI content. In addition to Meta, other large US tech companies such as Microsoft, Google and OpenAI have decided on measures to label AI-generated content.
Thanks to advances in AI technology, it’s becoming easier and easier to use software to create deceptively real-looking images – and now audio and videos. This can also be used to manipulate public opinion.