AI Leaders Are Sounding the Alarm. Who’s Listening?

AI is advancing so quickly that even its own creators are warning it could outrun our ability to control it. So why are Trump and Republicans siding with the industry against government oversight?

Over the weekend, Anthropic CEO Dario Amodei made a stunning admission: artificial intelligence is advancing so fast that, unless it slows down, it could threaten humanity itself. Other industry leaders, including Sam Altman and Elon Musk, quickly agreed.

That’s the good news. The people building AI now openly acknowledge that it poses real and immediate dangers.

The most alarming near-term danger is what researchers call recursive self-improvement: AI systems that can build better versions of themselves, each generation faster and more capable than the last, until the technology outruns our ability to control it. It is hard to overstate that risk.

If AI reaches that level soon, other crises will follow. Millions of jobs could vanish almost overnight. Machines could begin to stand in for the things that make us human: creativity, problem-solving, even friendship. We would face a national security crisis, an economic crisis, and a spiritual crisis all at once.

It’s a good thing AI leaders are finally waking up to the precipice they’ve brought us to. The problem is that the agreement ends there.

Amodei says we need government rules and approval processes now. But other companies, OpenAI and Elon Musk’s xAI among them, are spending millions to elect candidates who promise to block AI regulation. OpenAI is part of a coalition pushing Congress to bar every state from regulating AI at all.

The President has sided with them. This week he waved away the danger, saying what matters is “beating China.” In Congress, Republicans have drawn a line against any regulation. Speaker Mike Johnson, a favorite of much of the AI industry, said flatly this week that it is up to the companies, not the government, to protect the nation.

That is absurd. It shows how completely market fundamentalism and corporate loyalty have blinded the Republican Party to reality.

AI companies have no incentive to protect us. They answer to investors, not the public, and those investors are pushing them to reach recursive self-improvement as fast as possible, because whoever gets there first will make the most money. Any company that slows down on its own risks losing its investors and, with them, its business. So they won’t protect us. They can’t. As long as money is the only incentive, they will race to build the machine that could destroy us.

The only way to slow that race is one set of rules that applies to every company, and only government can write them. That is what government does best: set rules that preserve an industry’s benefits and control its dangers. We did it with nuclear power, with prescription drugs, and with workplace safety. We made sure nuclear energy didn’t destroy us, medicines didn’t poison us, and factories didn’t kill their workers. Public oversight of dangerous industries didn’t used to divide the parties. It united them.

Today a dangerous alliance is forming between the AI industry, with Amodei the notable exception, and the Republican Party. It threatens to leave the industry’s own warnings unheeded.

Warnings matter only if we act on them. So far, most of the AI industry wants to sound the alarm about what it has built while clinging to the fantasy that it can manage the danger on its own. In the end, that belief may be more dangerous than the technology itself.