Trump Cancels Signing of Executive Order Granting Oversight of A.I. Models

President Trump canceled the signing of an executive order that would have granted the federal government oversight to evaluate new artificial intelligence models before their public release [1]. The last-minute cancellation followed the President's stated concerns over 'certain aspects' of the order [1]. The order would have reversed the administration's hands-off policy on AI, granting the Office of the National Cyber Director and other agencies two months to develop an evaluation process [1]. Its goal was to identify security vulnerabilities in AI models to help protect banks, utilities, and other sensitive industries from cyberattacks [1]. The White House had also proposed that major AI companies voluntarily share their models 14 to 90 days before a public release [1]. The cancellation occurred hours before a scheduled Oval Office event, to which the CEOs of OpenAI, Google, Anthropic, Meta, and Microsoft had been invited just 24 hours earlier [1]. Some companies had already flown executives to Washington for the signing [1]. President Trump explained the decision, stating, 'I think it gets in the way of — you know, we’re leading China, we’re leading everybody, and I don’t want to do anything that’s going to get in the way of that lead' [1]. The scramble over the order highlighted divisions within the White House on how to manage a technology that affects both national security and the U.S. competitive race [1]. Officials had pushed for oversight stemming from fears that AI was becoming too powerful and could pose a future security risk, concerns amplified last month after startup Anthropic announced a new model capable of finding software vulnerabilities [1]. The debate over AI oversight intersects with broader discussions about the technology's societal and environmental impacts. The computational demands of large language models (LLMs) like GPT-4 carry a significant carbon footprint, with training a single model emitting carbon dioxide equivalent to hundreds of cars driven annually [3]. Furthermore, research into the outputs of leading proprietary and open-source LLMs from companies like OpenAI, Google, Microsoft, and Meta shows that their generated texts can be highly similar to each other and distinct from human writing, raising questions about diversity and ethical standards in automated systems [4]. The proposed executive order sought to address one facet of risk by focusing on security vulnerabilities, but its cancellation leaves the regulatory landscape for a rapidly evolving and resource-intensive industry unresolved [1].