The UK Government’s Mythos AI Tests Separate Hype from Reality in Cybersecurity
- The UK government’s AI Security Institute has evaluated Anthropic’s Mythos Preview model, adding independent public verification to earlier reports of its cybersecurity capabilities.
- While Mythos Preview shows no significant differences from other recent frontier models in individual cybersecurity tasks, it excels at chaining these tasks into multistep attack series necessary to infiltrate some systems.
- The AISI’s evaluations align with previous tests showing steady improvement in AI model performance on Capture the Flag challenges since early 2023.
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.