US government now has pre-release access to AI models from five major labs for national security testing
Five AI labs just handed the US government pre-release access to their models. Here's why national security testing is reshaping AI governance.
Why it matters
Government access to pre-release AI models for national security testing represents a major shift in AI governance—blurring the line between public safety oversight and classified defense priorities, with implications for model release timelines and competitive dynamics among the five labs.
The key facts
5 to knowFive major labs signed agreements: Anthropic, OpenAI, Google DeepMind, Microsoft, xAI
Center for AI Standards and Innovation facilitating pre-release access
Models provided with reduced safety guardrails for classified environment testing
Stated rationale: cybersecurity risks and US-China tech competition
Expands prior agreements already in place with Anthropic and OpenAI
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: The US Department of Commerce is expanding its AI safety testing: Following Anthropic and OpenAI, Google Deepmind, Microsoft, and xAI have now signed agreements with the Center for AI Standards and Innovation. The companies provide models with reduced safety guardrails for testing in classified…