US President Donald Trump has issued an executive order on Tuesday establishing a voluntary system where AI developers will share advanced models with the government before making them public. The main aspect of the order allows companies like OpenAI, Google, or Anthropic to provide the government with access to their most potent models for a period of up to 30 days before their intended release.
This decision was prompted by concerns surrounding Anthropic’s Mythos model, which the AI startup has chosen not to release publicly due to its potential to reveal vulnerabilities in computer systems, including those of banks, governments, and hospitals. The compromise reached in the order involved a 30-day pre-release government access window, as opposed to the initial draft that suggested a 90-day period. On the other hand, tech companies had advocated for a shorter 14-day window.
The executive order signing follows a period of uncertainty, with the White House initially close to unveiling the measure, only to retract suddenly. Reports from Politico and other sources indicated that David Sacks, the Silicon Valley venture capitalist and Trump’s AI and crypto czar, cautioned the president that the order could impede innovation and harm the US in its AI competition with China, surprising White House staff who believed Sacks supported the order.
In a recent statement on X, Sacks emphasized that excessive regulation poses the most significant threat to innovation in America, stressing that prevailing in the AI race necessitates navigating bureaucratic obstacles imposed by state legislatures and Washington policymakers. Additionally, the order directs Treasury, the National Security Agency, and the CISA agency to establish an “AI cybersecurity clearinghouse” in collaboration with industry and critical infrastructure operators to coordinate vulnerability scans for software and prioritize patches.
President Trump rescinded an AI oversight order from the Biden administration on his return to the White House. The previous order from Biden in 2023 required AI companies to share safety test outcomes with the government and relied heavily on voluntary commitments, falling short of the expectations of many experts.
In contrast, the European Union’s AI Act, which came into effect in 2024, enforces binding regulations for high-risk AI systems, mandating transparency requirements and imposing obligations on safety testing and incident reporting for the most advanced models.
