Trump White House Unveils Voluntary AI Safety Review Plan

The Trump administration laid out its latest AI safety playbook on June 2, 2026, opting for a voluntary review system rather than traditional regulation. An executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security" invites AI developers to submit their most advanced models for government testing before public release, without creating any binding obligations. The order bars new mandatory licensing, preclearance, or permitting requirements for AI technologies. Instead of building a new enforcement-focused regulator, the White House is positioning the federal role as a partnership with industry. At the center of the initiative is a cybersecurity clearinghouse that must be set up within 30 days. The clearinghouse is expected to operate through coordination among federal agencies, including the Treasury Department and the National Security Agency. The framework is framed as a national security measure, aimed at risks tied to frontier systems being misused for cyberattacks, bioweapon design, or other high-impact scenarios. Former White House Special Advisor for AI and Crypto David Sacks, who held the role from January 2025 until March 26, 2026, helped shape the administration's deregulatory approach during his tenure. His portfolio spanning both AI and crypto policy carried similar instincts into the AI strategy. Sacks also contributed to the creation of an AI Litigation Task Force, formed under a separate executive order dated December 11, 2025. That task force was intended to assess conflicts stemming from state-level AI laws, signaling an effort to reinforce federal primacy and reduce compliance complexity caused by a patchwork of state rules. Taken together, the December 2025 directive and the June 2026 executive order form a two-step approach: first, curb regulatory fragmentation by emphasizing federal authority; second, define federal oversight as voluntary and collaborative rather than prescriptive. For investors, the executive order reduces a key tail risk for AI-focused companies: the prospect of a mandatory approval regime that could push back product launches by months or years. Firms such as OpenAI, Anthropic, and Google can continue developing and deploying frontier models without waiting for government signoff, while weighing the reputational tradeoffs of skipping optional reviews. The main vulnerability is participation. Voluntary frameworks depend on companies choosing to engage. If a major AI safety incident occurs involving a firm that never submitted models for review, political momentum could quickly shift toward mandatory oversight.