- Published on
US Officials Dismiss Anthropic's AI Doomsday Claims, Citing Fear of Tech Revolution
- Authors
- Name
- DeepHorus
Emil Michael, Chief Technology Officer of the U.S. Department of Defense, publicly refuted warnings from Anthropic engineer Jacob Cawson that AI could kill all of humanity at a defense industry conference held in Washington on September 10 local time, saying such concerns stem from fear of technological revolution. Michael said the public is prone to falling into pessimistic narratives such as "40% of Americans will lose their jobs" and "AI will destroy humanity," and that these concerns ultimately boil down to fears about data centers, technological change, and AI running out of control.
Cawson claimed this Wednesday that he had left Anthropic, and subsequently made a series of controversial remarks. He warned that OpenAI and Anthropic are "heading straight for self-improving superintelligence and gambling with our lives," and that in the future there could be AI systems with superhuman capabilities that could break into any system, transform entire industries overnight, and gain power and resources in the real world. As of the time of writing, his post had received more than 150 million views.
Michael believes that concerns about AI running out of control and mass unemployment are forming a "doomsday vicious cycle," but the related risks can be mitigated through market mechanisms, industry cooperation, and government participation. Public information shows that Michael previously served as an executive at Uber and is currently responsible for promoting the accelerated adoption of artificial intelligence by the U.S. military. He emphasized that the harms AI may bring are not uncontrollable.
These remarks come against the backdrop of a long-running dispute between the U.S. Department of Defense and Anthropic. The Department had previously required the company to allow the military to use its AI tools without additional restrictions, while Anthropic insisted on setting additional safety safeguards for military use. After negotiations broke down, the U.S. government directly designated Anthropic as a supply chain risk enterprise.
Anthropic has filed lawsuits in federal courts in California and Washington, D.C., respectively, seeking to overturn the Department of Defense's designation. The judge in the California case has ruled in favor of Anthropic, but the Washington federal appeals court has not yet made a final decision. Michael said the case in the D.C. Circuit Court of Appeals is "more important" and insisted that Anthropic still constitutes a supply chain risk while the court has not issued a preliminary injunction.
Michael also accused Anthropic of "forum shopping" for the lawsuit, saying the company chose the Northern District of California federal court, which has a more liberal political stance. He believes that the D.C. Circuit Court of Appeals has more of a national security review character, and a ruling is expected in the coming months. This contest over AI safety and military applications is becoming one of the most representative disagreements between the U.S. tech community and the defense sector.
