U.S. Defense Official Dismisses AI Doom Narratives Amid Safety Disputes with Anthropic

Summary of Thoughts on the IT Home Article

Okay, so I've got this piece from IT Home, dated September 11th. It's about Emil Michael, the CTO of the U.S. Department of Defense – they also call it the "Department of War," which is interesting in this context – responding to concerns raised by Jacob Coxon, a former Anthropic engineer. Coxon's warning is pretty dire: he's talking about AI potentially wiping out humanity.

Michael seems to be downplaying the alarm, though. He’s saying that the public is getting caught up in these pessimistic narratives, like AI causing massive job losses or total human extinction. His take is that it all boils down to these underlying fears about data centers, rapid technological change, and the idea of AI becoming completely uncontrollable.

He seems to think we're stuck in a "doomist vicious cycle" around AI's dangers, but that these risks can be managed. He mentions market forces, collaboration across the industry, and government involvement as potential solutions.

Now, this Coxon guy, he resigned from Anthropic this past Wednesday, and immediately went public with some pretty explosive statements. Apparently, his posts have gone viral, racking up over 150 million views.

Coxon's argument is that OpenAI and Anthropic are rushing headlong toward superintelligence, and they're essentially gambling with all of our lives. He paints a picture of future AI systems that could be superhuman, able to hack any system, reshape entire industries overnight, and seize real-world power and resources.

Michael, who was previously at Uber and is now leading the U.S. military's AI adoption efforts, doesn't see the AI threat as insurmountable. He stresses that these risks can be controlled through market mechanisms, collaboration, and government action.

This all comes against the backdrop of an ongoing dispute between the DoD and Anthropic. Apparently, the Pentagon wanted to be able to use Anthropic's AI tools without any additional restrictions, but Anthropic wanted to put in some extra safety measures for military applications. Negotiations broke down, and the U.S. government has now classified Anthropic as a supply chain risk.

Anthropic is fighting back, filing lawsuits in both California and Washington, D.C., seeking to overturn this designation. They've already had a win in California, but the federal appeals court in D.C. hasn't ruled yet.

Michael is calling the D.C. Circuit Court of Appeals case "more important" and is sticking to the line that Anthropic is a supply chain risk, even without a preliminary injunction from the court.

He's also accusing Anthropic of "forum shopping," suggesting they chose the Northern District of California federal court because it's politically more liberal. He believes the D.C. Circuit is better suited for national security reviews, and he's expecting a ruling in the next few months.

IT Home, September 11 news: Emil Michael, Chief Technology Officer of the U.S. Department of Defense (also referred to as the "Department of War"), responded on September 10 local time during a defense industry conference in Washington to warnings from Anthropic engineer Jacob Coxon that AI could wipe out humanity.

Michael stated that the public is prone to falling into pessimistic narratives such as "40% of Americans will lose their jobs" and "AI will destroy humanity." He argued that these concerns ultimately boil down to fears surrounding data centers, technological disruption, and AI running out of control.

He believes that anxieties over AI slipping out of control and causing mass unemployment are generating a "doom-loop," but that the associated risks can be mitigated through market mechanisms, industry collaboration, and government involvement.

Coxon announced this Wednesday that he had resigned from Anthropic and subsequently posted a series of controversial remarks. As of IT Home's publication, his post has surpassed 150 million views.

He asserted that OpenAI and Anthropic are "racing toward self-improving superintelligence and gambling with our lives." He warned that superhuman AI systems could emerge in the future capable of breaching any system, transforming entire industries overnight, and seizing real-world power and resources.

Public records indicate that Michael, a former Uber executive, is currently tasked with accelerating the U.S. military's adoption of artificial intelligence. He remarked that the potential harms of AI are not unmanageable, noting that beyond market mechanisms, risks can also be diminished through industry collaboration and government participation.

These comments come against the backdrop of an ongoing dispute between the U.S. Department of Defense and Anthropic. The Pentagon previously demanded that the company permit the military to use its AI tools without additional restrictions, while Anthropic insisted on implementing extra safety safeguards for military use. After talks broke down, the U.S. government designated the company as a supply chain risk.

Anthropic has filed lawsuits in federal courts in California and Washington, D.C., seeking to overturn the Defense Department's designation. A judge in the California case has ruled in Anthropic's favor, but the federal appeals court in Washington has yet to deliver a final ruling.

Michael called the case before the U.S. Court of Appeals for the D.C. Circuit "more important," maintaining that absent a preliminary injunction from the court, Anthropic continues to pose a supply chain risk.

Michael also accused Anthropic of "forum shopping" for the litigation, stating that the company chose the politically more liberal federal court in the Northern District of California. He contended that the D.C. Circuit is more attuned to national security reviews, and a decision is expected in the coming months.