In his return to the White House for a second term, President Donald Trump has repealed a 2023 executive order issued by his predecessor, Joe Biden. The 2023 order aimed to address the risks posed by AI to national security, public health, and consumer safety. Its revocation signals a major policy shift in the US government’s approach to AI oversight.
Biden’s executive order required developers of AI systems that could impact security or public welfare to submit safety test results to the federal government before public deployment. The measure, rooted in the Defence Production Act, also directed federal agencies to establish safety standards and mitigate risks, such as those related to cybersecurity, and chemical, biological, radiological, and nuclear threats. The National Institute of Standards and Technology (NIST) was tasked with creating guidance to help companies identify and address flaws in AI models, such as algorithmic biases.
The 2024 Republican Party platform had pledged to repeal the Biden-era order, describing it as an obstacle to innovation. Party leaders argued that the reporting requirements placed excessive burdens on developers and risked exposing proprietary information. Trump, during his campaign, had promised policies to support AI development grounded in free speech and economic growth, though specific details remain undisclosed.
The revoked order had been part of Biden’s broader push to implement safeguards for the growing AI industry, which has seen rapid advances in generative AI. These technologies, capable of producing text, images, and videos, have sparked concerns over potential job losses and other societal impacts.
Continued focus on AI safety standards
The repeal comes as global and domestic initiatives to enhance AI safety standards gain traction. In July 2024, the US Department of Commerce released new AI safety guidelines, following the Biden administration’s directives. These included documents from NIST outlining measures to counter adversarial risks to AI systems and draft guidance from the US AI Safety Institute to address emerging threats.
International collaboration has also intensified. In April 2024, the US and UK signed a memorandum of understanding to advance global AI safety efforts. The agreement aims to facilitate joint testing of AI models, personnel exchanges, and information sharing between the two nations, reflecting the shared urgency to address the challenges posed by AI’s rapid adoption.
Although Trump has rescinded Biden’s AI safety order, an executive order addressing the energy demands of advanced AI data centres remains in place. Issued by Biden last week, the directive calls for federal sites to host gigawatt-scale AI data centres and renewable energy facilities to meet the infrastructure needs of AI development. This policy highlights the increasing resource requirements of the technology sector as AI capabilities expand.
Read more: US Department of Commerce unveils new AI safety and security guidelines
