Researchers at the US company Calif have announced a new method for bypassing MacOS security layers using the Anthropic Claude Mythos neural network. By combining two distinct vulnerabilities and additional techniques, experts managed to access restricted memory segments on Mac devices. This is reported by Ixbt.com .
This type of attack is known as "privilege escalation." Although this method is not sufficient for full device control on its own, it poses a risk of total system compromise when combined with other vulnerabilities. Using Claude Mythos, they successfully bypassed Memory Integrity Enforcement (MIE), a technology Apple described as a "five-year engineering achievement."
According to Calif representatives, it took only five days to create the exploit code. However, they emphasized that the AI did not do all the work alone; the expertise of cybersecurity professionals played a crucial role in the success.
Apple stated that it has launched an investigation into the report and that security is their top priority. Calif plans to fully disclose the technical details of the attack once Apple has patched the identified vulnerabilities.
Read “Zamin” on Telegram!
