Microsoft, Google, and xAI to give early access of next-gen AI models to US Govt

In a significant move earlier this week, the Trump administration announced its plan to oversee the deployment of artificial intelligence (AI) models before they are made available to the general public. Major technology players such as Microsoft, Google DeepMind, and Elon Musk's xAI have reportedly agreed to provide the US government with early access to their next-generation AI models for national security evaluations. The National Institute of Standards and Technology (NIST) issued a press release confirming that the Centre for AI Standards and Innovation (CAISI) has entered into an agreement with these leading companies. This partnership will empower the institute to perform pre-deployment assessments and targeted research to better understand the capabilities of advanced AI technologies and enhance AI security measures. This agreement is particularly timely, given the rising cybersecurity concerns surrounding AI models, including Anthropic's Claude Mythos, which has ignited debates about the potential for misuse of powerful AI systems. In response, governments and regulators around the world are increasingly advocating for stronger safeguards and heightened transparency in AI development and usage. Chris Fall, the Director of CAISI, emphasized the importance of independent and rigorous evaluation of frontier AI technologies, stating, "Understanding frontier AI and its implications for national security is crucial. These expanded collaborations with industry help us to scale our efforts in the public interest during this critical time." Beyond pre-launch evaluations, the initiative will also encompass research and assessments of AI models after they have been deployed. Meanwhile, the White House is working on assembling a team of experts tasked with providing guidance on how the government can effectively review AI models. Discussions are underway with tech leaders from companies such as Google and OpenAI to help develop a framework for AI security assessments, safety standards, and oversight protocols.