Instagram to Label AI Posts, Not Filter Them, as Detection Gets Harder
Updated
Updated · The Verge · Jul 10
Instagram to Label AI Posts, Not Filter Them, as Detection Gets Harder
3 articles · Updated · The Verge · Jul 10
Summary
Adam Mosseri said Instagram will label AI-generated content rather than remove or broadly filter it, arguing users who dislike it should be able to train their feeds away from it.
Mosseri drew a line between sorting and banning, saying some users may want feeds full of AI content while the platform should answer whether a post is probably AI, uncertain, or not AI.
Detection remains a core problem: Mosseri said identifying AI content is hard and could become impossible as models improve, making labels for camera-captured "real media" potentially more practical.
Instagram is still trying to curb spammy AI posts even as Meta expands AI tools such as Muse Spark, which lets users place tagged people into generated images.
That rollout has already drawn criticism from the National Center on Sexual Exploitation, which warned the feature could enable harassment, sexual abuse, exploitation and identity fraud.