UK Gov’s Mythos AI Tests Help Separate Cybersecurity Threat from Hype
- The UK government’s AI Security Institute (AISI) has conducted an initial evaluation of Anthropic‘s Mythos Preview model, adding independent public verification to Anthropic’s claims of the model’s striking capability in cybersecurity tasks.
- AISI’s findings indicate that while Mythos isn’t significantly different from other recent frontier models in individual cybersecurity-related tasks, it could set itself apart through its ability to effectively chain these tasks into multi-step series necessary for full system infiltration.
- Since early 2023, AISI has been testing various AI models, including Anthropic’s Mythos Preview, through Capture the Flag challenges. The performance of subsequent models has steadily improved, with Mythos Preview completing north of 85 percent of “Apprentice” level tasks.
Originally published at arstechnica.com. Curated by AI Maestro.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.