Microsoft Google xAI preview AI models for US security review
Technology
Microsoft, Google, and Elon Musk's xAI are giving the US government a sneak peek at their latest AI models.
The idea is to let officials check for national security risks before these powerful tools go live for everyone.
This move was announced by the Center for AI Standards and Innovation (CAISI), part of the US Department of Commerce.
CAISI completed more than 40 evaluations
This approach follows similar agreements in 2024 with OpenAI and Anthropic.
CAISI has already completed more than 40 early-stage model evaluations (sometimes even without safety filters) to spot potential issues.
Meanwhile, the Pentagon has reached agreements with seven AI companies to test out technology on secure networks.