The Trump administration announced new collaborations with major technology firms, including Microsoft, Google DeepMind, and Elon Musk's xAI, aimed at advancing research in artificial intelligence (AI) and enhancing national security measures. The initiative is facilitated by the Center for AI Standards and Innovation (CAISI), part of the Commerce Department's National Institute of Standards and Technology.
These agreements expand upon earlier partnerships, focusing on pre-deployment evaluations and specific research regarding the capabilities and security of frontier AI technologies. CAISI Director Chris Fall emphasized the importance of rigorous measurement science in understanding the implications of advanced AI, particularly concerning national security. The partnerships aim to strengthen information-sharing among stakeholders, encourage voluntary improvements in AI products, and provide government officials with a clearer understanding of AI capabilities and international competition.
As part of the agreements, developers will provide CAISI with AI models that may have altered safety features for evaluation purposes, addressing national security risks. Evaluators from various government agencies will contribute feedback through the TRAINS Taskforce, which is dedicated to AI-related national security concerns. The agreements also allow for testing in classified environments and are designed to adapt to ongoing advancements in AI technology.
Microsoft's Chief Responsible AI Officer, Natasha Crampton, stated that the collaboration is essential for building trust in advanced AI systems. She noted that thorough testing is critical for understanding system performance and identifying potential risks, such as AI-driven cyberattacks. Microsoft has also signed a similar agreement with the United Kingdom's AI Security Institute to govern AI testing and evaluation.