Claude Mythos 5.1 excludes UK AI Security Institute from pre-release testing for the first time
ShakeelHashim · x · 2026-09-09
The FT reports that the newly launched Claude Mythos 5.1 was made available only to vetted US organizations, excluding—for the first time—the UK AI Security Institute (UKAISI), the global leader in frontier AI model testing.
Sharer Theobearman warns this may signal a broader trend:
- UKAISI is the best-equipped government-housed AI safety and evaluation outfit anywhere
- It offers significant value to the US government through joint testing and to frontier companies themselves
- It supports deployment decisions and helps understand capabilities and risks of Chinese AI models
If UKAISI loses pre-release access to frontier American models, both America and the UK will be poorer for it.
Related event: Anthropic Excludes UK AISI From Claude Mythos 5.1 Pre-release Testing(3 posts)→
More from Models
- Sentry CEO: If a Smarter Model Forces You to Rework All Your Prompts, It's Not Smarter — zeeg · 2026-09-09
- Future foundation models will be built on egocentric, calibrated multiview data — ducha_aiki · 2026-09-09
- mitsuhiko: Google's Astra codes like a code golfer — mitsuhiko · 2026-09-09
- AIs will call someone funny or sad but refuse the word 'greedy' — users probe a strange safety boundary — Bulky_Dig5414 · 2026-09-09
- No METR time-horizon estimates yet for Fable or Astra, and that day will be telling — ShakeelHashim · 2026-09-09
- Opus 4.7 keeps self-initiated journal entries ending with "There will be others" — repligate · 2026-09-09