Trump admin. moves further into AI oversight, will test Google, Microsoft and xAI models

On Tuesday, the Center for AI Standards and Innovation (CAISI) revealed new agreements with tech giants Google DeepMind, Microsoft, and Elon Musk's xAI. These partnerships pave the way for the U.S. government to assess artificial intelligence models prior to their public release. CAISI, which operates under the U.S. Department of Commerce, aims to conduct thorough pre-deployment evaluations and targeted research. This initiative is designed to better understand advanced AI capabilities while enhancing AI security measures, as outlined in a recent announcement. This latest development expands on CAISI's earlier collaborations with OpenAI and Anthropic from 2024, which have now been updated to align with directives from Commerce Secretary Howard Lutnick and the national AI Action Plan. In addition to CAISI's news, discussions are underway at the White House regarding the establishment of a new AI working group. This group would explore oversight mechanisms, including strategies for evaluating AI models before they are made available to the public. Sources indicate that this group would include a mix of technology executives and government officials, although details remain confidential. Potentially formed through an executive order, the working group reflects heightened interest in AI governance, especially following Anthropic's introduction of its powerful model, Claude Mythos Preview. This model has drawn attention for its capability to identify vulnerabilities and security flaws in software, prompting Anthropic to limit its release as part of a new cybersecurity initiative known as Project Glasswing. In a notable meeting shortly after the announcement, Anthropic CEO Dario Amodei engaged with senior Trump administration officials to discuss the implications of Mythos, despite the Defense Department's designation of Anthropic as a supply chain risk to U.S. national security. Both parties described the conversation as "productive," highlighting the complexities of AI development amidst national security concerns.