An updated Claude 3.5 Sonnet underwent the first-ever joint pre-deployment evaluation by the U.S. and U.K. AI safety bodies.

The AI company is in talks with federal agencies, its head of global affairs said in an interview with FedScoop.

Only some of the 16 companies explicitly said they’d stick to the AI pledges made to the current administration.

The former top innovator at the Government Accountability Office, Taka Ariga, is hopeful that the unit will remain “vanguard of experimentation and exploration” for Congress and others.

The AI Safety Institute will have access to new models before and following their releases under the new testing and evaluation pacts.

Sponsors of the legislation to authorize a national research resource for artificial intelligence told FedScoop they were optimistic about its path forward.

The pilot brings together research resources from multiple federal and industry partners and will serve as a “proof of concept” for the full-scale project, according to NSF.

The commitments include committing to pre-release security testing for AI models and forming insider threat safeguards and cybersecurity investments focused on unreleased and proprietary model weights.