Instagram won't filter AI—but will let you build an AI-only feed, says Mosseri

Instagram's leader Adam Mosseri has clarified his stance on AI-generated content: the platform will label it clearly, but won't automatically remove it from users' feeds or provide filters to exclude it entirely.

Labeling Over Filtering#

Speaking on Lenny Rachitsky's podcast, Mosseri distinguished between identifying AI material and censoring it. Rather than scrubbing posts from the feed, he said Instagram should tell users whether content is AI-generated and let them decide how much exposure they want. Those who prefer AI creations, he suggested, should be able to "have a feed that's just AI town."

The Detection Challenge#

Mosseri acknowledged that spotting AI work will only grow harder as generative models advance, potentially degrading Instagram's detection abilities over time. He proposed a more nuanced approach: letting users query whether a specific post is likely AI-made, with the system responding that it "probably is, or we're not sure, or it's definitely not." As an alternative, he floated the idea of marking human-captured camera content instead—a strategy he'd mentioned in an earlier discussion about fingerprinting authentic media.

Broader Platform Tensions#

Though Mosseri indicated Instagram needs to combat spammy automated AI content, the platform continues integrating AI tools. Meta's Muse Spark image generator now allows users to insert tagged individuals into their AI creations with a single action. Haley McNamara, executive director of the National Center on Sexual Exploitation, warned that this feature "creates obvious and foreseeable opportunities for exploitation, sexual abuse, harassment, and identity fraud."