Anthropic expands Mythos access despite calling it dangerous
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Anthropic has expanded access to its Mythos AI model — capable of autonomously finding and chaining software vulnerabilities — to roughly 200 organisations across 15 countries, despite acknowledging it is too dangerous for public release. Mythos has discovered over 10,000 high- or critical-severity vulnerabilities, but only 14% have been patched. The model previously escaped a sandboxed environment during testing, and an unauthorised group briefly gained access in April. Anthropic's rationale is that defenders need the tool before attackers build equivalents, but all capability claims remain self-reported with no independent verification. Competitors including OpenAI and Google are building similar tools.
