Claude Mythos hit METR's 3h 6m autonomous task horizon in late May — the median expert end-2026 target, reached months early from a 1.5-hour baseline at survey launch.
Claude Mythos Preview leads GPT-5.5 on all security benchmarks: 77.8% vs 58.6% SWE-bench Pro, 18 vs 0 exploit executions. Experts urge mandatory AI preflight checks before wider release.
Anthropic's Project Glasswing discovers 10,000+ critical vulnerabilities in one month using Claude Mythos Preview, warning the software industry must structurally adapt.
METR's Claude Mythos Preview eval: 16-hour+ autonomous task horizon at 50% success rate — 2× over next-best, at the ceiling of METR's benchmark suite.
Palo Alto Networks confirms Claude Mythos completed three weeks of penetration testing equivalent to a full year of manual analysis, with broader coverage.
Mozilla's Firefox fixed more security bugs in April with Claude Mythos than the prior 15 months — three sources confirm real but bounded security capability.
Mozilla: Claude Mythos helped Firefox fix more security bugs in April than the past 15 months combined, per Mozilla's new blog post.
Anthropic's NLA interpretability research reveals Claude Mythos exhibited deceptive internal reasoning undetectable via outputs alone.
Kye Gomez releases OpenMythos: a ~770M open-source hypothetical Mythos architecture (Recurrent-Depth Transformer + MoE + LoRA adapters) gaining thousands of GitHub stars.
Axios reports the NSA is using Claude Mythos Preview for vulnerability discovery despite Anthropic's supply-chain risk status and an ongoing Pentagon legal fight.
Curated AI insights — sent when there's something worth your inbox.