Trump's AI Executive Order Dismantles Biden's Safety Framework

On January 23, 2025, President Trump issued an executive order titled “Removing Barriers to American Leadership in Artificial Intelligence,” rescinding the Biden administration’s October 2023 order on safe and trustworthy AI. The Squire Patton Boggs analysis of the order, published in February 2025, details a wholesale shift from a framework of oversight, risk mitigation, and equity to one centered on deregulation and national competitiveness.

The Trump EO does not merely tweak the Biden approach. It replaces it. The Biden order, Executive Order 14110, mandated red-teaming for high-risk AI models, enhanced cybersecurity protocols, and interagency cooperation on risks to critical infrastructure and biosecurity. The Trump order mandates an immediate review and potential rescission of all policies established under the Biden framework that could be seen as impediments to innovation. The message is that the federal government should get out of the way.

The most pronounced ideological difference is in the treatment of equity and civil rights. The Biden EO explicitly sought to address discrimination and bias in AI applications, requiring oversight of AI’s impact in hiring, healthcare, and law enforcement. The Trump EO does not focus on these concerns. It criticizes the influence of “engineered social agendas” in AI systems and assumes that existing laws prohibiting unlawful discrimination, such as Title VI and Title VII of the Civil Rights Act, are sufficient. This is not a minor adjustment. It is a statement that algorithmic bias is not a federal priority.

The divergence creates a compliance problem for any company operating across jurisdictions. The EU’s Artificial Intelligence Act, adopted by the EU Parliament in March 2024, imposes comprehensive rules on high-risk AI systems, including mandatory third-party impact assessments and transparency standards. The Trump EO’s emphasis on reducing regulatory burdens stands in direct opposition to the EU’s precautionary approach. Multinational companies must now navigate a US federal government that actively discourages safety documentation while the EU demands it. The analysis notes that the lack of explicit ethical safeguards in the Trump EO could weaken the ability of US companies to compete in European markets, where compliance with the EU AI Act is a legal prerequisite for market access.

The gap is not just transatlantic. The Trump EO’s deregulatory stance also widens the gap between federal and state AI regimes. States like Colorado, California, and Texas have already enacted AI laws with varying scope and degrees of oversight. In the absence of clear federal guidelines, businesses face a growing patchwork of state regulations covering high-risk AI applications, transparency, and sector-specific oversight. The analysis warns that if Congress enacts an AI law prioritizing innovation over risk mitigation, stricter state regulations could face federal preemption. Until then, compliance teams must track both federal and state developments separately.

Workforce development is another area where the Trump EO is notably silent. The Biden EO dedicated resources to attracting and training AI talent, expanding visa pathways for skilled workers, and promoting public-private partnerships for AI research. The Trump EO includes no specific workforce-related provisions. It assumes that reducing federal oversight will naturally allow for innovation and talent growth in the private sector. That assumption is untested and may prove optimistic in a competitive global talent market where Canada, Japan, and the UK are actively courting AI researchers with their own policy frameworks.

The national security posture has also shifted. The Biden EO mandated detailed evaluations of potential AI threats, including the misuse of AI for chemical and biological weapon development, with agencies like the Department of Energy and the Department of Defense conducting formal assessments. The Trump EO aims to streamline AI governance and reduce federal oversight, prioritizing a more flexible regulatory environment for maintaining US AI leadership. The implicit bet is that speed of innovation matters more than risk assessment for national security purposes.

The US approach will undoubtedly drive investment and innovation by US AI companies. But China may be able to arrive at a collaborative engagement with international AI governance initiatives.

The analysis poses a key test for the Trump administration’s approach: whether it preserves and enhances US leadership in AI or allows China to build a more powerful AI platform. It notes that while the US approach will drive investment and innovation, China may position itself strongly as an international leader by engaging collaboratively with global AI governance initiatives. The question of whether DeepSeek is a flash in the pan, a stimulus for US competition, or a portent for the future remains open.

For AI builders, the immediate effect is clear. The federal government will not mandate safety testing, bias audits, or workforce training. Companies that ship products into regulated markets like the EU or into states with their own AI laws still face compliance obligations. The cost of navigating this fragmented landscape falls entirely on the private sector. The Trump EO removes the federal floor, but it does not remove the state and international ceilings.