Federal agencies would strengthen AI cybersecurity and model testing

Federal agencies must prioritize cyber defense, create an AI cybersecurity clearinghouse, review existing grant funding, and expand certain cybersecurity hiring pathways. The directions cover national-security systems, Department of War systems, civilian federal systems, state and local authorities, and critical infrastructure such as hospitals, banks, and utilities. The clearinghouse would operate through voluntary cooperation with AI companies and critical-infrastructure operators. The grant review identifies funding that is already available and relevant; it does not itself award new money. Agencies must act under existing legal authority and only when appropriations are available. Most cybersecurity and grant-review actions must begin within 30 days of June 2, 2026. The hiring-pathway expansion must occur within 60 days.

The named agencies must develop a process to test advanced AI models and a voluntary framework for developers to share qualifying models with the Federal Government and trusted partners. The framework covers models designated as “covered frontier models” because of their advanced cyber capabilities, including models still under development. Developers may ask the Federal Government whether a model qualifies and may provide access for up to 30 days before release to other trusted partners. Access must protect confidential information, cybersecurity, insider-risk controls, intellectual property, permitted use, and nondisclosure obligations. The order does not authorize mandatory licensing, preclearance, or permits for developing, publishing, releasing, or distributing AI models. The agencies must develop the benchmarking process and voluntary framework within 60 days of June 2, 2026.

The Attorney General must prioritize enforcement of applicable federal criminal laws against people who use AI to illegally access or damage computer systems or support other crimes. The priority covers unauthorized access to public or private information systems, unlawful data collection, computer damage, and AI agents used to obtain information for criminal purposes. This direction prioritizes enforcement of existing criminal laws and does not itself create a new criminal offense. Implementation remains subject to applicable law and available appropriations. The order gives no separate deadline for this enforcement priority.