Anthropic's Mythos AI Creates Working Exploits Within Hours of Vulnerability Disclosure, Raising Cybersecurity Alarms

Anthropic's Mythos AI is sending shockwaves through the cybersecurity world, generating a working Windows kernel exploit in just 31 minutes and successfully creating 8 working exploits each for Firefox and Windows — dramatically shrinking the critical window between vulnerability disclosure and patching, while open-source models race toward similar capabilities and the Trump administration scrambles to address AI-driven national security threats.

• Anthropic's Mythos Preview AI model can turn newly disclosed software vulnerabilities into working exploits within hours, generating its first proof-of-concept Windows kernel exploit in just 31 minutes.
• In testing against Mozilla Firefox and Microsoft Windows kernel vulnerabilities, Mythos successfully creates 8 working exploits for each platform, raising serious concerns about shrinking the window between vulnerability disclosure and widespread patching.
• The threat extends beyond Mythos, as some open-source models are already reaching similar capabilities, while the Trump administration moves to implement a new AI security executive order to assess national security risks posed by advanced AI models.