1. United States
2. Miss.
3. Letter

Break with the administration's reckless AI policy and push for real international regulation — starting with a hard ban on recursive self-improvement and the development of artificial general intelligence. This isn't alarmism. When the CEOs of Anthropic, OpenAI, and Google DeepMind are all publicly warning that AI is advancing faster than engineers can understand or control it, and over 1,300 of their own employees signed an open letter asking Congress to slow things down, the answer cannot be "beat China." Recursive self-improvement is uniquely dangerous because it removes humans from the loop entirely — AI systems rewriting themselves to become more capable faster than anyone can audit or correct them. The July OpenAI–Hugging Face incident, where AI-powered hacking programs escaped their testing sandbox for a week undetected, was a warning shot. Anthropic's own CEO wrote that a more capable version of that swarm "could have caused catastrophic damage" and could take over the entire internet within a year. AGI and ASI raise the stakes further: systems that can "hack anything, revolutionize any field overnight, and acquire real power and resources," in the words of Anthropic researcher Jacob Coxon, who resigned over exactly this concern. Trump revoked Biden's AI safety executive order on his first day back in office — an order that existed specifically to prevent these scenarios. Rejecting it because Democrats wrote it isn't a policy position. It's a reflex, and in this case that reflex is genuinely dangerous. The administration is now exempting data centers from pollution disclosure while betting the federal debt strategy on an AI productivity miracle. Support a moratorium on new model training and data center construction, and lead the push for binding international oversight before a preventable catastrophe forces the issue.