Interview with Instagram's Adam Mosseri about using AI to deal with bullying on the platform, reassessing its core incentives, breaking up big tech, and more
Katy Steinmetz / TIME :
Context & Ripple Effects
This interview lands mid-arc in Adam Mosseri's run as Instagram's public face. A month earlier, a [[a:942312|CNBC profile reported he had privately pushed to remove fake news and demote Breitbart in News Feed]] — so his willingness to discuss "reassessing core incentives" and even breaking up big tech reads as a continuation, not a one-off. The bullying question also previews the incentive work that followed: within months he was framing Project Daisy, the removal of public Like counts, as a way to "depressurize the app".
What makes the AI-for-bullying pledge worth tracking is how it ages against later coverage. By 2021 Mosseri was answering questions about keeping users safe at scale, and by 2024 he was attributing account-locking moderation failures to human moderator "mistakes" rather than AI systems — meaning the automation he pitched here became part of a system whose failures he would later pin on the humans around it.
First-order effects
- Users targeted by bullying are the immediate beneficiaries if Instagram deploys AI to detect and intervene in harassing interactions, shifting enforcement from reactive reports toward proactive detection.
- Mosseri publicly entering the break-up-big-tech debate puts him, as Facebook's most visible product executive, on record ahead of the antitrust scrutiny that dominates his subsequent Q&As about future regulation.
Second-order effects
- If Instagram demotes engagement metrics and leans on AI enforcement, rival platforms face pressure to match both moves — the Like-hiding experiment that followed set a template competitors had to decide whether to copy.
- Automating anti-bullying work reduces dependence on human review capacity, which becomes a double-edged dependency when moderation errors later trigger user lockouts across Instagram and Threads.
Third-order effects
- The pattern here — executives proposing structural self-correction (incentive redesign, AI enforcement) while acknowledging regulators may force bigger changes — points toward platforms treating trust as infrastructure they must engineer, not just moderate.
- If AI-mediated enforcement becomes the default at scale, accountability debates shift from what content platforms allow to how their automated systems decide, foreshadowing the regulator interest Mosseri discusses in later interviews.
The trend: Platform chiefs are moving from defending engagement-maximizing design toward AI-driven safety enforcement and incentive redesign, partly to get ahead of regulators weighing structural breakups.