OpenAI and Meta trade AI agent privacy promises — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI unveiled its "Dots" AI agent at DevDay, with CEO Sam Altman declaring the company wants to "set a new standard for privacy in frontier AI" while taking veiled shots at Meta's Muse
- Meta's Muse gained 600,000 daily US daily active users within weeks (per Apptopia) and topped App Store charts, but a security researcher quickly exposed a zero-day vulnerability, and 404 Media reported last-minute issues including one that could have let users access Meta's internal databases
- Meta itself can still access user data stored on Muse's "isolated linux computer" VMs, and Muse defaults to allowing Meta to train models on user inputs — undercutting Mark Zuckerberg's claim Muse was "built from the ground up for privacy and security"
- OpenAI head of app platform Glen Coates said the company is "in a different position to Meta" because Meta "doesn't have an AI product that has 1.2 billion users," while pitching enterprises "stronger controls" and zero data retention policy options
- Dots is only available on ChatGPT's $100-and-up subscription tiers, limiting both its user base and the number of privacy incidents that can surface to test those promises
- The Verge's Allison Johnson described feeling uncomfortable typing her bank information into Muse when prompted as part of a task — illustrating the broader tension that AI agents inherently demand deeply personal data to function
Why it matters: Privacy has become the primary battleground for AI agent adoption, but Meta's Muse stumbles — a zero-day exploit, pre-launch security flaws, the agent handing out a user's address, and Meta itself retaining data access — show these promises break fast. Users are the ones absorbing the risk of handing over bank info and private messages, while companies pocket the goodwill of claiming they're the safer choice.
Ask SkimNews
