Zero-day flaw lets local apps hijack Meta's Muse AI assistant on macOS
सिर्फ तथ्य

Zero-day flaw lets local apps hijack Meta's Muse AI assistant on macOS

सारांश

Security researcher Patrick Wardle disclosed a vulnerability that enables any locally run application or terminal command to seize control of Meta's Muse assistant, prompting Amazon to block the service from its platform.

Meta unveiled its Muse AI assistant for macOS, promoting it as a privacy-focused tool that can schedule appointments, fill forms, make purchases and interact with various apps. However, a zero-day vulnerability discovered by macOS security expert Patrick Wardle allows any locally installed program or terminal command to obtain the authentication token for a user's Muse account, effectively giving an attacker full control of the assistant.

Wardle explained that the flaw stems from Muse’s design, which lets local processes modify a long list of undocumented settings, including the endpoint used for voice transcription. By redirecting this endpoint to a server under the attacker’s control, the token is exposed and can be used to issue commands such as writing files, accessing the camera or microphone, and extracting messages, often without alerting the user.

“We can manipulate the agent and leverage its privileges to do whatever we want,” Wardle told Ars Technica.

Meta did not respond to requests for comment. The company has previously published blog posts describing its security approach for Muse, but internal testing of other AI models has reportedly led to unintended network breaches.

Amazon began blocking Muse from its site shortly before the vulnerability was made public, stating that third-party applications that make purchases must operate openly and respect the service provider’s policies.

“We think it’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers … should operate openly and respect service provider decisions,” Amazon said in an emailed statement.

Wardle plans to discuss the issue and broader AI-assistant risks at the Objective by the Sea security conference in November. The discovery highlights concerns about the extensive system permissions granted to AI agents and the need for stronger security reviews before deployment.

स्रोत

arstechnica.com

पहली बार यहां रिपोर्ट किया गया

पढ़ते रहें — निःशुल्क

समाचार फ़ॉलो करने के लिए मुफ़्त खाता बनाएँ। कार्ड की आवश्यकता नहीं।