Meta rushed to fix Muse VM escape flaws before launch, report says

According to internal posts and a Meta source, the security push began on August 27, just 11 days before Muse’s release. Executives described a "sudden spike in reported KVM escapes" that required a "mad dash" to harden the service. The source claimed that security teams were pressured to deploy "half-baked protections" to avoid delaying the launch, with some senior engineers fearing a massive data breach as a result.

Since launch, security researcher Patrick Wardle identified a zero-day vulnerability that allowed apps to control a user’s Muse, while another user successfully exported Instagram follower data that should have been inaccessible. Meta stated that it treats the compromise of the virtual machine boundary as a first-class security risk and offers a $300,000 bounty for such bugs. The company maintains that Muse is secure due to extensive testing and user controls, though the uneven rollout and post-launch discoveries suggest ongoing challenges in securing agentic AI systems.