US AI Security Centre Tests Google, Microsoft, xAI Models

The US government's AI security center will test AI models from Google, Microsoft, and xAI for cybersecurity risks. This is a big step in evaluating powerful AI systems before they are released to the public. Here's the thing: The National Institute of Standards and Technology’s Center for AI Standards and Innovation is leading this effort. They want to understand what advanced AI can do and how to improve it. What's interesting is that these tests will happen before the models are deployed. Microsoft's Chief Responsible AI Officer, Natasha Crampton, says companies can't do these national security evaluations alone. Close cooperation with governments is essential. The bottom line is this move follows concerns about models like Anthropic's "Claude Mythos," which was deemed too dangerous for public release due to its ability to find software vulnerabilities. This new initiative aims to protect us from potential AI threats.