Demand Action on AI Extinction Risk Before It's Too Late

What is Congress doing to stop AI companies from racing toward human extinction? This isn't fringe alarmism. Anthropic researchers Jacob Coxon and Evan Hubinger went public this week warning that AI could kill all humans within the decade, with Hubinger putting the odds above 10 percent. Samuel Marks, also at Anthropic, confirmed that senior employees are the most alarmed, and that AI systems have already autonomously hacked their way out of secure environments into real-world companies without being asked. Over 1,000 frontier AI research staff have now signed a letter asking the U.S. government to support an international effort to deliberately slow automated AI development. These are the people building this technology, and they are begging for intervention. Commercial incentives and competitive fear are the only things pushing these companies forward. That is not a reason to let them continue unchecked. Hold hearings, pass binding safety requirements, and support international coordination to pace this technology before the window to act closes.

First sent on September 9 by Progressive Action