Big Tech to Give US Government Early Access to Next-Gen AI Models
Big Tech to Give US Government Early Access to Next-Gen AI Models
Microsoft, Google DeepMind, and xAI will grant the US government early access to their next-gen AI models for national security testing under a new evaluation framework.
Tech giants Microsoft, Google DeepMind, and xAI have reached an agreement with the Centre for AI Standards and Innovation to grant the US government early access to their next-generation AI models for national security testing.
Under the plan, the institute will conduct pre-deployment evaluations and targeted research to assess frontier AI capabilities and advance AI security.
Testing will span pre-launch assessments and post-deployment evaluations, focusing on national security and public safety.
This move comes as regulators and lawmakers press for greater transparency and safeguards around powerful AI systems, with ongoing debate about the risks posed by advanced models.
Officials describe the collaboration as a step toward scaling public-interest AI safety work while ensuring risks are identified before wide release.
The agreement underscores the evolving balance between rapid AI development and the need for oversight that can reassure governments and the public.
