A former AI researcher just lit Washington, DC on fire with a warning that could not be ignored.
Ted Cruz read every word of it — and his response was not what anyone expected from the man who chairs the committee that oversees the industry.
Now the warning set off a firestorm no one can ignore.
The Warning That Rattled Capitol Hill
Jacob Coxon resigned from Anthropic and posted a thread on X that racked up over 150 million views in a matter of days.
Coxon, who spent three years doing pretraining research at both OpenAI and Anthropic, did not mince words. “They are racing straight to self-improving superintelligence and gambling with our lives,” he wrote.
He went further. “The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon stated. That sentence landed like a freight train across the political spectrum.
Evan Hubinger, the alignment science lead at Anthropic, then went public to back Coxon up. Hubinger said he personally estimates greater than 10 percent odds of extinction within a decade and acknowledged that Anthropic does not yet have a concrete plan to control superintelligent systems. That is not a fringe blogger. That is a senior safety researcher at one of the most powerful AI labs on earth admitting the lab does not know how to stop what it is building.
Anthropic CEO Dario Amodei has separately warned that AI could wipe out half of all entry-level white-collar jobs and drive unemployment to 10 to 20 percent within one to five years. The people inside this industry are sounding alarms that go well beyond anything Congress has been willing to take seriously — until now, apparently.
Cruz Calls It Scary As Hell — Then Explains Why He Won’t Stop It
US Senator Ted Cruz (R-TX), chairman of the Senate Commerce Committee, sat down for an interview with Politico’s Dasha Burns for the program “The Conversation,” and what he said deserves a careful read.
Cruz acknowledged reading Coxon’s thread directly. “Listen, I read that same thread from that Anthropic engineer, and that was scary as hell,” he said.
He also said something that no one who follows his record on technology regulation would have expected: “We need substantially more regulation when it comes to AI.”
But here is where it gets complicated. Cruz immediately pivoted to the argument that has defined his position all along — Communist China is not slowing down, and the United States cannot afford to either.
When Dasha Burns noted that a “bipartisan cohort of voices” continues to press to “slow this thing down,” Cruz was blunt. “Here’s the problem — we can’t. That is an impossible endeavor, because you know who’s not pausing? China.”
“The Chinese are putting no guardrails. They’re going full speed ahead. Most tech leaders estimate the United States is about six months ahead,” he continued. “We could have government step in and just shut down the AI industry in America, and you know what would happen? China would catch up in six months, and soar ahead, and we wouldn’t have the tools to defend ourselves.”
Asked whether Congress can keep up with the pace of the industry, Cruz said: “It would be a terrible outcome if we stifle U.S. innovation, and China pulls ahead. That would be dangerous for Americans.”
And then came the line that will follow this interview everywhere it gets shared. Burns asked whether the logic meant that if AI is going to kill everyone, it should at least be American AI doing it. Cruz leaned into it. “Listen, I’ve actually joked — and this is somewhat tongue-in-cheek, but it’s not entirely — if there are going to be killer robots, I’d rather they be American killer robots and not Chinese killer robots.”
He did add one caveat: “If there are things that have real risks of catastrophic damage, there needs to be rules and guardrails.”
What the Senate Commerce Committee Is Actually Preparing to Do
Cruz told Politico that the Senate Commerce Committee could take up AI regulation after the chamber returns to Washington.
But there is a catch. Several committee members have said they remain in the dark about the specific details of what Cruz has planned for that markup. The gap between his public urgency and what his colleagues actually know is drawing attention from technology policy advocates and industry lobbyists watching Capitol Hill closely.
Cruz has previously said that any markup is “designed to move legislation that has a real chance of passing into law,” and that he vets proposals based on “what bipartisan agreement and consensus can be reached.” His staff has signaled that federal rules should apply only in “truly novel circumstances” where current statutes fall short — think catastrophic risks, deepfakes, and chatbots interacting with children.
That is a narrower scope than the language Cruz used in the Politico interview, and the tension between those two positions is real.
The honest question sitting underneath all of this is whether Washington can actually regulate something that moves this fast. Big Tech companies building these systems are not waiting for Congress to figure out its jurisdiction. They are shipping products, training models, and expanding computational capacity at a pace that makes the standard legislative timeline look like a slow-motion replay.
And those same Big Tech companies — the ones now racing toward systems that their own researchers say could kill everyone — are the same corporations that spent years censoring conservatives on COVID policy, questions about the 2020 election, and the events of January 6. The companies asking for the public’s trust on existential risk are the same companies that decided which political opinions American citizens were allowed to read. That context does not disappear just because the subject has shifted from speech to superintelligence.
Anthropic CEO Dario Amodei’s warning about mass white-collar job displacement is not a projection from a critic. It is an admission from inside the industry itself. When the man running one of these companies tells the public that AI could eliminate half of all entry-level white-collar jobs within five years, that is not an argument for accelerating without guardrails. It is an argument for asking hard questions before the damage is done.
Cruz is right that Communist China is not pausing. Beijing is pushing ahead with no concern for the safety questions that Coxon and Hubinger raised publicly, and a unilateral American slowdown that hands the lead to an authoritarian surveillance state would be its own kind of catastrophe. Whoever wins the AI race will embed their values into systems that shape how billions of people live. If Communist China wins that race, the result is global surveillance and control by a regime with no regard for individual liberty.
But “China won’t stop” is not a complete answer to “our own researchers think this could kill everyone.” Those two facts can both be true at the same time, and Congress is going to have to figure out how to hold them together.
The Senate Commerce Committee markup, whenever Cruz decides to reveal its actual contents, will be one of the most consequential votes on technology policy in years. The fact that members of his own committee do not know what is in it yet is either a sign of careful deliberation or a sign that the answers are harder than the public statements suggest.
Probably both.
Sources: Politico, “The Conversation” with Dasha Burns (September 10, 2026); Mediaite, “Ted Cruz on AI: ‘I’d Rather They Be American Killer Robots!'” (September 10, 2026); Joe.My.God., “Ted Cruz: AI Researcher’s Warning Is ‘Scary As Hell'” (September 10, 2026); TechCrunch, “‘Gambling with our lives’: Anthropic researcher quits, warns against self-improving AI” (September 9, 2026); The National News, “‘AI could kill us all’: Anthropic researcher Jacob Coxon quits with warning about powerful AI models” (September 10, 2026); congress.net, “Ted Cruz Calls Viral AI Researcher’s Warning ‘Scary As Hell’ Amid Push For Regulation” (September 10, 2026)
