The White House moved to tighten federal cyber defenses around artificial intelligence, unveiling an executive order that directs agencies to craft a voluntary pre-release review for “frontier” AI models and to stand up an AI cybersecurity clearinghouse. The order, a shift from the administration’s previously light-touch posture, stops short of mandatory licensing but compels closer coordination between government and industry as emerging systems show superior capability to find and exploit software flaws.
Under the plan, developers would offer the government access to advanced models 30 days before broader release, while security agencies led by CISA, NSA and NIST set thresholds and classified benchmarks for when reviews apply. The administration also called for binding directives within 30 days to harden critical systems and for expanded AI-enabled defensive services to state, local and infrastructure operators. Treasury, working with cyber agencies, will coordinate vulnerability disclosure and remediation through a new clearinghouse. OMB is tasked with surfacing grants for AI cyber tools, and OPM will widen cyber hiring channels under the Tech Force initiative.
Industry groups praised the voluntary, phased approach as supportive of innovation, while civil-liberties advocates warned against backdoor licensing and arbitrary enforcement. Some security professionals criticized classified benchmarking and limited early-access pilots as barriers to broader defender use. Supporters said the clearinghouse could lift resilience if it translates insights into practical fixes at scale; skeptics cautioned that without staffing and context, more vulnerability data might overwhelm organizations rather than protect them.
Related article:
