Meta's Muse Is Better at Surveilling Than Helping Me — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Meta's Muse was downloaded over 900,000 times in its first week per Sensor Tower, and is cross-promoted across Instagram, WhatsApp, and Messenger with a default avatar described as a beige cross between an Ewok and a Labubu
- Muse automatically opts users into AI model training, with Meta saying data is "sanitized" to remove identifying information before training, though the article notes the process is unclear and data may include information from connected sources like email or bank accounts
- Muse stores user preferences in a "Memory" document but offers no toggle to fully disable the feature—users can only request deletion via chat or manually edit the file, and the app's "ideas" tab constantly prompts linking bank accounts, inboxes, and personal documents
- Calli Schroeder of the Electronic Privacy Information Center flagged the opt-out model as undermining trust, citing Meta's recent decision to opt all adult Instagram users into an AI deepfake tool and pull it days later as evidence of "manipulative tactics"
- Margaret Mitchell of Hugging Face warned that agentic tools like Muse disengage users and may cause "cognitive degradation" as people outsource taste and preference decisions, while Rory Mir of the EFF noted that AI chat windows are effectively users putting information directly into Meta servers
- Meta plans to release "confidential" virtual machine versions of Muse later this year that "cryptographically and verifiably prevent" the company from accessing data inside, and the agent's web browsing can still indirectly shape Instagram ads via tracking algorithms
Why it matters: Meta's 900,000 first-week Muse downloads represent a massive new pool of users automatically opted into AI training data collection, with the app actively soliciting bank logins, inboxes, and personal documents—repeating a rollout-first pattern that drew backlash when Meta recently opted all adult Instagram users into an AI deepfake tool before pulling it days later.
Ask SkimNews
