Trump Has No Worries About AI Extinction. His Country's Top AI Researchers Do.
As insiders at OpenAI and Anthropic warn about AI that could rewrite its own code beyond human control, the U.S. president says extinction risk isn't on his list.
Key points
- President Trump said Thursday he has no concerns about AI posing an extinction risk to humanity.
- Anthropic researcher Jacob Coxon resigned Tuesday, saying AI companies are "gambling with our lives."
- OpenAI chief scientist Jakub Pachocki said Saturday he expects AI progress to reach recursive self-improvement, a state where AI upgrades itself without human input.
- Two bipartisan bills are moving through Congress: one to govern how advanced AI models are deployed, one to pause that development until safety rules exist.
- Trump said the U.S. Leads China in AI development by roughly a year, framing that race as the real risk.
President Trump got a direct question Thursday: does he worry AI could wipe out humanity? His answer was four words. "No, I don't have any."
The concern he does have is losing. "If we don't win AI, we're going to be put in a very bad position," he told reporters, pointing to China. He put the U.S. Lead at about a year, which he called significant.
Why are researchers sounding the alarm right now?
Several people who actually build these systems stepped forward this week with warnings that go far beyond competitive anxiety.
Jacob Coxon, a researcher at Anthropic, the company behind the Claude family of AI assistants, quit his job Tuesday. His stated reason: he believes Anthropic and OpenAI are "gambling with our lives." He said researchers inside these labs believe the technology could "kill us all by the end of the decade." Other employees at both labs echoed that view publicly within days. AI2Day has been covering Coxon's warnings since 9 September, including what his resignation actually means for the rest of us.
The specific fear centres on recursive self-improvement, or RSI. Today's AI is built and improved by humans. RSI describes a future state where an AI system rewrites its own code, getting smarter faster than people can understand or stop it. Jasmine Wang, an OpenAI researcher working on alignment, which means keeping powerful AI pointed at what humans actually want, wrote Wednesday that it is "hard to overstate how dangerous speeding towards RSI is."
OpenAI's chief scientist Jakub Pachocki went further on Saturday. He said he has a "strong expectation" that AI progress will reach that self-improving stage. "I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence," he said. We covered Pachocki's warning in detail on 9 September, and it's worth noting his company has since sought legal cover to slow the race down without running into antitrust law, as we reported the following day.
What is Congress doing about it?
Legislators from both parties are pushing for rules, though they disagree on how far to go.
In July, Representatives Jay Obernolte and Lori Trahan announced the FRONTIER Act, a bill to create a legal framework for deploying the most advanced AI models. Trahan wrote on Wednesday: "Safety researchers are resigning, powerful AI models are breaking out of their labs, and companies are racing ahead anyway. It's past time for Congress to get off the sidelines."
Senator Bernie Sanders and Representative Greg Casar went further with the Ban Artificial Superintelligence Act, introduced earlier this month, which would freeze development of the most advanced AI systems until the federal government sets safety rules.
Two very different responses to the same set of facts. What matters most here isn't which side sounds scarier. It's which side writes the rules first, and right now that is genuinely unresolved.
