Meta bolsters Muse safety warning after security vulnerability found, The Information reports
The flaw could let an attacker with code already on a Mac redirect transcription and reach a user’s dedicated virtual machine, Meta said.
- Security researcher Patrick Wardle discovered a zero-day vulnerability in Meta's Muse AI that allowed attackers with local code access to seize control of the agent, capture voice prompts, and manipulate sensitive device functions.
- Meta initially classified the flaw as a "SEV-2," its third-highest severity level, after internal reports indicated it could expose user virtual machines containing personal emails and files. The company patched the vulnerability shortly after receiving the report.
- Developers Peter James and Jonny Saunders independently coaxed Muse into revealing its internal filesystem, exposing plain-text Markdown and JSON files detailing how the agent processes requests, manages data, and connects to services like Gmail.
- Meta spokesperson Daniel Roberts defended the filesystem access as "intended behavior," describing Muse as a "computer in the cloud." The company denied the incident represents a security breach, asserting that exporting data provides no privileged access.
- Despite the security concerns, Muse has rapidly climbed app charts with about 2.8 million downloads in its first two weeks. Wardle advised users to keep permissions limited and avoid installing AI agents that request broad access to device resources.
28 Articles
28 Articles
‘Please don't install': Security researcher sounds alarm over Meta's Muse AI for MacBook owners
Installing Meta's new Muse AI assistant on your MacBook could open the door to cybercriminals. Security researcher Patrick Wardle discovered a zero-day vulnerability in the new tool. These are the worst vulnerabilities in any software because "zero-day" means hackers already know about the flaw and are using it to hack into devices. It literally refers to the fact that there are zero days to fix the issue before bad actors discover it, so it's a…
A security vulnerability has been discovered in Meta’s personal AI agent, ‘Muse,’ which allows attackers to access a dedicated cloud space where users’ emails and files are stored. Meta has taken steps to strengthen warnings displayed when users access dangerous websites. Citing an internal Meta incident report on the 25th (local time), the U.S. IT news outlet The Information reported that
(San Francisco = Yonhap News) Correspondent Kwon Young-jeon = A security vulnerability has been discovered in Meta's new artificial intelligence (AI) agent 'Muse,' which is gaining sensational popularity.
Patrick Wardle, Apple's security specialist, discovered a vulnerability in Meta's new AI agent.The attack could take advantage of the permissions the user had already given the assistant to access files, camera and connected services.The researcher anticipated Clarín that they found problems in other agents.
Channel News Asia
Meta bolsters Muse safety warning after security vulnerability found, The Information reports
Sept 25 : Meta Platforms is adding a clearer safety warning within Muse after a security researcher discovered a vulnerability in the AI agent that could let an attacker access a user's sensitive personal information, the Information reported on Friday.The flaw, reported by an outside researcher through Meta'
Coverage Details
Bias Distribution
- 46% of the sources are Center, 45% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
