Story perspectives

Britain’s AI Security Institute gets early access to frontier models

By Drooid · Oct 13, 2026, 10:11 AM UTC · How we work

Summary

Story summary

Britain’s AI Security Institute will receive early access to frontier AI models for safety testing.
AI Minister Kanishka Narayan said the institute will evaluate Gemini 4 Argon and Anthropic’s Mythos 5.1.
U.S. President Donald Trump’s administration reversed a ban, permitting U.S. developers to share new models with the institute.
The institute’s voluntary agreements let it assess models before deployment and help shape emerging international safety standards.

1 of 1

Story summary

Britain’s AI Security Institute will receive early access to frontier AI models for safety testing.
AI Minister Kanishka Narayan said the institute will evaluate Gemini 4 Argon and Anthropic’s Mythos 5.1.
U.S. President Donald Trump’s administration reversed a ban, permitting U.S. developers to share new models with the institute.
The institute’s voluntary agreements let it assess models before deployment and help shape emerging international safety standards.