What happened
Anthropic unveiled Claude Mythos, an AI model Anthropic states surpasses human capabilities in cybersecurity and hacking tasks. Anthropic reported 'Mythos Preview' red team testing identified thousands of high-severity vulnerabilities, including a 27-year-old flaw in critical systems. Access to Mythos is restricted to over 40 organisations, including Amazon Web Services, Apple, Microsoft, Google, Nvidia, Broadcom, and Crowdstrike, via Project Glasswing, an initiative to strengthen software resilience against Mythos itself.
Why it matters
The ability of AI models to locate critical flaws without extensive supervision, as demonstrated by Mythos finding a 27-year-old vulnerability, increases the speed and scale of exploit discovery. Security architects and platform engineers face new challenges from these AI-accelerated threat capabilities. Financial institutions and governments, including Canadian finance ministers and the Bank of England, express serious concern, prompting discussions at the IMF and EU regarding potential undermining of financial system security.
Subscribe for Weekly Updates
Stay ahead with our weekly AI and tech briefings, delivered every Tuesday.
