美國攜手Google、微軟和xAI進行模型測試，加強人工智慧安全防護。

The United States Department of Commerce is now going to test new artificial intelligence models from Google, Microsoft, and xAI before these tools are released to the public. These companies have agreed to let the government check their AI systems through the Center for AI Standards and Innovation, also known as CAISI. This is an important step because AI is becoming a bigger part of our lives, and the government wants to make sure these new technologies are safe and reliable. The agreement builds on earlier deals with other AI companies like OpenAI and Anthropic, which also allowed their models to be tested for safety and security.

The testing will include looking at how these AI tools work, how safe they are, and how they can be improved. Google's most popular AI tool is Gemini, a chatbot that is used in many Google products. Microsoft has CoPilot, which helps people with writing and other tasks. xAI, a company connected to Elon Musk, has Grok, another chatbot that has gotten a lot of attention. CAISI has already tested 40 different AI tools, including some that have not been released to the public yet. They did not say which models were stopped from being released, but the testing is meant to make sure only safe and useful tools are shared with people.

Microsoft said in a blog post that it already tests its AI models, but working with the government is important for national security and public safety. Google did not comment on the new testing, and xAI did not respond to questions. This new way of working with AI companies is different from how things were done before. The Trump administration wanted fewer rules for AI, but now the government is being more careful. With the military using AI more and new claims about powerful AI models, the White House is changing its approach. They are talking to AI companies and making sure safety is a top priority. This shows that the US wants to lead in AI technology, but also wants to protect people from possible risks.