Meta to Label AI-Generated Content and Adjust Moderation Policy Starting May

Meta will provide AI-generated content with a Made with AI label from May. The tech giant said in February that such labels were coming, but did not yet indicate when that would happen. The company is also adjusting its moderation policy regarding manipulated media.

Meta’s policy will apply to Facebook, Instagram and Threads from May the tech giant wrote in a blog post on Friday. The company will from then on provide ‘images, videos and audio’ with a Made with AI label. When uploading content, users can indicate whether they have used AI. Meta says it can also independently detect ‘indicators’ of AI content with a detection system based on ‘industry standards’, although the company provides few concrete details about this.

Meta announced in February that it would start labeling AI-generated images later this year. The company was also talking about videos and audio at the time. The tech giant did not say exactly when it would start labeling AI content. Meta now says that this will happen from May onwards. The company has long labeled content generated with Meta’s own AI tools, but this will also apply to content generated by other services. Meta said in February that it was already working with other AI companies on technical standards for identifying AI content.

The tech giant is making further changes to its policy on manipulated media. Meta does that after feedback from its Oversight Board. Meta’s old policy dates back to 2020 and only applied to deepfakes that put words in people’s mouths. The policy therefore did not provide sufficient coverage for the wide range of manipulated media and AI content that is being distributed today, the Meta regulator concluded. An example of this was a manipulated video of US President Joe Biden. It wrongly suggested that he had behaved inappropriately, but was not technically prohibited under the old policy.

The new policy will therefore apply to more types of content. In addition, content that poses a ‘high risk’ of deception will be provided with an extra prominent label with context. Furthermore, from July onwards, AI-generated content will no longer be automatically removed if it does not violate Meta’s other terms and conditions. From then on, content will only be removed in the ‘most risky circumstances’.

Bron: Meta
2024-04-05 18:24:49
#Meta #platforms #provide #AIgenerated #media #labels