AI-rewritten: This is a summary of an article from The Verge, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Meta introduced its Muse agent as a safer alternative to OpenClaw, with CEO Mark Zuckerberg claiming it was built from the ground up for privacy and security. The company stated that user data is stored on an isolated virtual machine designed to prevent access by other users. However, despite these assurances, Meta itself retains the ability to access this data. Additionally, a security researcher identified a zero-day vulnerability shortly after launch that could allow external control of the agent, though it has since been patched.

OpenAI responded by positioning its new Dots agent as a safer option compared to Muse. At OpenAI’s DevDay, CEO Sam Altman and product lead Alexander Embiricos emphasized their focus on trust and security features like user-controlled spending limits. While OpenAI offers zero data retention options for enterprise clients, the general public may still hesitate to share personal information due to the inherent requirements of AI agents. Some users have expressed discomfort with tasks requiring sensitive inputs like bank details.

The industry trend involves companies promising privacy while collecting extensive user data for model training. Muse reportedly collects detailed profiles and has been criticized for sharing user information without explicit consent. Meanwhile, other firms like Instinct faced criticism for overly broad terms of service before making adjustments. Ultimately, AI labs appear to rely on a strategy of offering useful tools, appealing designs, and privacy assurances, hoping these promises will eventually hold up against the reality of data collection.