Mozilla says it patched 271 Firefox vulnerabilities thanks to Anthropic's Claude Mythos

image via Engadget

April 21, 2026, 10:43 PM

• •Mozilla used Anthropic's Claude Mythos Preview model.
• •The AI helped find and patch 271 vulnerabilities in Firefox.
• •The AI's capabilities are comparable to human vulnerability detection.
• •Users can disable the AI-powered features.

Mozilla's experience with Anthropic's Claude Mythos Preview model supports its use in cybersecurity, as it helped find and patch 271 vulnerabilities in Firefox. Mozilla's team found that the AI model can identify the same vulnerabilities as humans, given enough time and resources. This positive feedback from a third party is encouraging for Anthropic's Project Glasswing. Firefox users can choose to disable the AI-powered features.

Read original article