What happened
Meta has removed a new AI feature on Instagram that allowed users to modify photos from public accounts. As reported by TechCrunch, the tool let anyone generate images by @-mentioning a public account to reference its photos, without notifying the account owner.
The feature was part of Meta's new Muse Image generator. Following immediate user backlash, Meta announced in a blog post it was pulling the tool, saying it “missed the mark.” The decision came after scrutiny from users and talent agencies.
How the room's reading it
The backlash was swift and centred on consent. Privacy advocates and creators immediately flagged the feature as an obvious vector for abuse, allowing anyone to use public photos without permission or notification. The consensus on developer forums and X was that Meta had shipped a tool with predictable—and harmful—misuse cases baked in from the start.
The involvement of talent agencies like CAA underscored the commercial and reputational risks for public figures. Many practitioners are reading this as a classic case of a major lab prioritising a feature launch over basic ethical design and user safety, only to walk it back when the public reaction proved too negative to ignore.
Sailfish's take
This wasn't a simple misstep — it was a failure to ask the most basic question before shipping: how will people abuse this? We've seen this play out for years with generative tools. Building without a clear, explicit, and opt-in consent model for user data is no longer a defensible position for any team.
The lesson here for builders is simple. Consent isn't a feature you patch in after launch, and privacy isn't an edge case. If your product uses public data in a new way, the default must be 'no'. We'd advise any team working on similar tools to build their consent and safety guardrails before writing a single line of model code.