Meta Shifts to Labeling AI Deepfakes Over Removing Them
Meta announces a new policy to label AI-generated deepfakes instead of taking them down, responding to Oversight Board criticism and upcoming global elections. The social giant relies on industry AI indicators or user disclosure to apply these new badges.
Meta expands its rules for handling AI-generated content by introducing a "Made with AI" label for deepfakes starting next month. The company applies this badge only when it detects industry-standard AI image indicators or when the uploader explicitly discloses the synthetic nature of the media. This adjustment means that AI content lacking these specific markers escapes labeling entirely.
Instead of removing manipulated media, Meta shifts its strategy to prioritize transparency and additional context. The social network stops taking down content solely based on its current manipulated video policy in July, favoring a "more labels, fewer takedowns" playbook across platforms like Facebook and Instagram. This approach aims to balance the mitigation of deceptive media with the protection of free speech.
This policy change arrives as Meta faces intense legal pressure from the European Union's Digital Services Act and prepares for major global elections. With the European Parliament elections in June and the U.S. presidential election in November, Meta is proactively adjusting its moderation tactics to satisfy regional regulations. The update also directly responds to recent criticism from the company's independent Oversight Board regarding its handling of manipulated media.