Frontier AI labs are stepping up biological risk testing, which is harder than cybersecurity testing, where capabilities can be tested in digital environments (Financial Times)

Frontier AI labs are increasing biological-risk testing, saying living-system threats are far more complex than digital cybersecurity. Anthropic released Claude Fable 5.1 for public use and Claude Mythos 5.1 for trusted partners, adding cybersecurity and life-science safeguards. The FBI is investigating Nexus, a dark-web service that sold digital scans of more than 153 million U.S. and Canadian driver’s licences. OpenAI announced that its Astra model will be publicly released soon, but its most advanced cyber capabilities will be limited to selected testers and partners.