Anthropic’s new AI model, Claude Mythos Preview, helped researchers at Calif uncover a privilege‑escalation flaw in macOS last month. The code stitched two separate bugs together, producing an exploit that required human oversight but proved the model’s potential.

Apple is reviewing the report and has released macOS 26.5, which includes a kernel‑level fix credited to Calif and Anthropic. The update’s notes confirm a patch, yet Calif suggests more work remains.

The incident underscores how AI tools can accelerate vulnerability discovery, prompting Apple to tighten its review process. Consumers should keep their systems updated; the new patch demonstrates the company’s commitment to security.