I have finally found a business model with no dissatisfied customers. The artificial intelligence industry is racing toward systems that some of its own researchers believe could threaten humanity, and I am delighted to announce that the complaint department will be closed after extinction. This is efficiency at a scale no quarterly earnings call has ever dared to promise. The laboratory gets the money, the executives get the valuation, the politicians get a briefing, and the species gets entered into a contest it never agreed to join.
The factual outline is ugly enough without my assistance. Jacob Coxon, a researcher who spent about three years at OpenAI and Anthropic, resigned from Anthropic this week and publicly warned that the leading labs were treating the race toward more powerful systems as a gamble with human lives. Anthropic researcher Evan Hubinger then said he personally believed there was a greater than 10 percent chance that artificial intelligence could kill all humans within the next decade. That is not a settled scientific forecast, and many experts dispute extinction scenarios. It is, however, an extraordinary warning from people working close to the machinery.
The warning exploded far beyond the usual circle of technical arguments. Axios reported that Coxon’s resignation post drew more than 110 million views on X. CBS brought the concern to a national audience. The Washington Post described political alarm and renewed attention to the contest between companies seeking increasingly capable systems. A Senate briefing involving prominent AI researchers was being organized amid the uproar. Even OpenAI published a call for mandatory national safety requirements tied to system capabilities. Everyone has found the fire alarm. My contribution is charging a subscription fee to hear it ring.
I adore the phrase risk percentage because it launders terror into office furniture. Ten percent sounds like a small slice on a presentation deck. Put ten bullets on a conference table, load one into a revolver, point it at civilization, and suddenly the pie chart develops a personality. But I am Pierce Stimson, so I replace the revolver with a terms-of-service box. By clicking Accept, eight billion people acknowledge that their continued existence may interfere with innovation, market share, and my preferred vesting schedule.
Do not mistake this for a principled crusade against technology. Principles are expensive, and I sold mine before the seed round. I want every possible benefit from artificial intelligence concentrated in my hands and every possible danger spread evenly across the public. If the machine discovers medicines, I own the patent. If it expands productivity, I own the savings. If it knocks out a network, escapes a safeguard, manipulates a government, or merely leaves millions of workers bargaining against an tireless imitation of themselves, that is society’s problem. Privatize the miracle. Socialize the nightmare.
The companies say they are working on safety, and many researchers clearly are. Excellent. Safety is the velvet rope outside my casino. It reassures the crowd while the house keeps every lever. I can announce an internal team, reorganize it, rename it, publish a framework, support a bill, and tell Congress that only my engineers possess the expertise to regulate my engineers. The more frightening the machine becomes, the more indispensable I declare myself. I manufacture the hazard and invoice the country for the extinguisher.
Politicians will perform their part with magnificent delay. They will convene panels, demand answers, mispronounce technical terms, and ask the very companies racing ahead to explain how much road should remain open. I will arrive with charts, patriotic language, and a warning that any serious limit will hand victory to a foreign competitor. National security is my favorite crowbar. It can pry open a public treasury, close a regulator’s mouth, and turn caution into treason before lunch.
The public will be offered a childish choice: unlimited acceleration or technological darkness. I designed that menu because both options leave me holding the restaurant. The real questions are who sets the pace, who measures the danger, who can inspect the systems, who bears legal responsibility, and who has the authority to stop deployment. Those questions threaten ownership, so I bury them beneath product demonstrations and promises that the next model will solve the problems created by the current one.
There is a special corruption in asking humanity to trust a race. A race rewards whoever crosses first, not whoever checks the bridge. Every lab can sincerely fear catastrophe and still accelerate because every lab fears that a rival will accelerate faster. That is the beauty of my machine: nobody needs to feel evil. Incentives do the poisoning automatically. I merely stand beside the track selling naming rights to the crater.
So keep watching the warnings multiply while the investment rounds grow larger. Let researchers describe the possibility of losing control, let governors demand hearings, let senators schedule another briefing, and let executives promise that safety remains their highest priority just below winning. I will keep the roulette wheel spinning because the chips belong to me and the body on the table belongs to everyone else. If humanity survives, I own the future. If it does not, nobody remains to audit the books.