CAISSI officials to report directly to White House under Trump administration

CAISSI: CAISSI, the Center for Advancing Innovation and Standards for Super Intelligence, is a research unit at NIST within the Department of Commerce that serves as the primary government point of contact for testing and evaluating advanced AI models, developing voluntary standards, and assessing security risks. It works with industry on model evaluations and coordinates across federal agencies on AI capabilities and vulnerabilities. In this news, the Trump administration is restructuring CAISSI so its senior officials report directly to the White House to address longstanding resource issues and streamline its role in the administration's AI testing framework. White House: The White House serves as the central executive office managing presidential policy, administration priorities, and interagency coordination. It is implementing a plan for CAISSI officials to report directly to it amid confusion with the Commerce Department over agency management and its integration into broader AI initiatives. Leo Schwartz: Leo Schwartz is a reporter at The Information focused on technology and politics. He broke the scoop detailing the CAISSI restructure, including debates over agency management and personnel changes under the Trump administration. Commerce Department: The US Department of Commerce oversees NIST and houses CAISSI as part of its efforts to advance innovation and standards in emerging technologies. CAISSI currently operates within it but is undergoing a restructure that involves direct White House oversight to resolve operational challenges. AI Governance: CAISSI contributes to the Trump administration's AI Action Plan through taskings on measurement science, interagency coordination, voluntary standards, and evaluations of advanced AI systems. Agency Restructure: The planned move of CAISSI leadership aims to resolve confusion between the White House and Commerce Department on how to manage the agency's role in federal AI model testing.