Trump privately signs scaled-back AI executive order that cuts down a voluntary review period

On Tuesday, President Donald Trump enacted a revised executive order that grants artificial intelligence and technology firms a limited timeframe to allow the federal government to voluntarily assess their AI models. This decision follows multiple discussions between the government and major AI players, including Anthropic, regarding cybersecurity risks. In April, Anthropic made headlines by announcing a pause on releasing its model, Claude Mythos, due to fears that its capabilities could expose software and cybersecurity vulnerabilities. The startup, which is in the process of preparing for an initial public offering, indicated last week that it would soon make Mythos-level models available to its clients. The updated executive order allows developers of advanced AI models to offer the federal government access to their technologies up to 30 days prior to public launch. This marks a significant change from an earlier draft of the order, which suggested a review period of up to 90 days. Importantly, the new directive clarifies that these assessments are entirely voluntary. Trump has previously expressed concerns about maintaining the United States' competitive edge in the AI sector, particularly against China. "We're leading China. We're leading everybody," he stated to reporters on May 21, emphasizing his reluctance to implement measures that could hinder this advantage.