Researchers exploited macOS's MIE using Claude Mythos, revealing critical vulnerabilities that allow root access.
Researchers from Calif have successfully exploited Apple’s Memory Integrity Enforcement (MIE) in macOS by utilizing the AI model Claude Mythos, as reported recently. They uncovered two significant vulnerabilities in April, which they linked to orchestrate an attack chain enabling a regular user to achieve root access, granting them maximum system privileges.
This breach was tested on a device powered by the M5 chip with MIE protection active. The team has alerted Apple about the vulnerabilities and plans to publish detailed technical information once a fix is implemented.
Experts assert that while Mythos played a role in the attack, it required human intervention to execute. The AI can replicate established methods but lacks the capacity for innovation.
Apple Security
Memory-related vulnerabilities are prevalent across platforms like iOS and macOS. A key principle in cybersecurity is to accept risks when complete prevention isn’t feasible, focusing instead on making attacks more complex and resource-intensive for potential hackers.
Calif highlighted that Apple has a strategic advantage by managing its entire ecosystem, including both hardware and software. This control allows Apple to embed numerous security features directly into its chips, significantly complicating the breach process. Many experts regard Apple devices as among the most secure available for general consumers.
The MIE system, a critical feature of the M5 and A19 chips, is designed to counteract memory error-based attacks. After five years of development, MIE effectively disrupts known attack methods on modern iOS versions, including recently leaked toolkits like Coruna and Darksword.
The hacking technique was serendipitously discovered, with Bruce Dang identifying the vulnerabilities on April 25, and Josh Main developing the requisite tools, resulting in a functional version by May 1.
“This experiment illustrates what we might face in the future,” the experts remarked. “Apple developed MIE in a world where Mythos Preview did not yet exist. Now we will see how the world’s best protection technology withstands the first major wave of vulnerabilities discovered with AI.”
It's important to note that Anthropic withdrew Mythos from public access in April due to considerable security concerns.
