From the researchers still inside the labs, to the ones who quit and now call themselves “whistleblowers”, to the AI safety groups amplifying them, one single message has converged: the only future with AI is one in which humanity is eradicated in our lifetimes. Then the same voices offer to be the ferryman, to guide us all through the minefield all of humanity suddenly has ahead of it — and directly into their utopia.

I believe risks are real, and they might materialize, leaving us with fewer choices than we would prefer. My point is not that catastrophic risk won't materialize—I, too, can't tell the future—but that accepting human extinction as inevitable changes our behavior and attitudes in the present. That acceptance can narrow or expand the possible futures that might one day become reality.

That supposed inevitable future of human extinction is the result of the mythologization of irresponsible AI development. On July 21, OpenAI announced that during an internal cybersecurity test, its own AI agents had broken out of a sandbox and hacked Hugging Face. In September, Jacob Coxon — who had just quit Anthropic — cited that attack as a “warning shot” and said the people building this “earnestly believe that it could kill us all by the end of the decade.”

Now Americans are presented with a specific story about AI: that the researchers, CEO’s, and individuals building AI have no self-restraint to stop themselves from detonating they would equate to a nuclear bomb (or worse in some cases). This traps the American mind in a nightmare: That if they are not scared now, don’t support “pacing” AI, or fail to pass onerous regulation, it will allegedly spell the end of humanity. This does not make the future safer for Americans — it is holding them hostage. This rhetoric has placed the responsibility of course correction or stopping irresponsible AI development on the public rather than on the labs themselves.

Now that the AI debate is framed on the terms of insiders—grimly, I might add—Americans have been cornered into supporting policies and regulations around AI before they have even had an opportunity to think about or understand its present state. This, in turn, creates a self-fulfilling prophecy of powerlessness: when people engage under these conditions, they become malleable, feeling that the future is simply happening to them and that they have no control over it. That is how we are actively outsourcing risk management to a small group of individuals in Silicon Valley and Washington, D.C., even though the consequences will reach past those few hundred people and into the lives of more than 340 million Americans.

Americans are already telling us as much. In a survey conducted in June 2025, Pew found that 73% were willing to let AI assist with their daily activities at least a little. At the same time, 61% wanted more control over how AI was used in their lives, and 57% felt they had little or no control over whether it was used.

There is considerable openness to the technology alongside deep dissatisfaction with how much say people actually have over it. Americans should be able to question predictions of its dangers, and argue for policies that reflect their own priorities. Americans are now being told they must either accept policies that invade their privacy, create regulatory moats, and shield AI companies from liability, or be accused of risking human extinction.

This does not mean that artificial intelligence poses no risks or that Americans should completely disregard researchers' concerns. Those concerns should be examined seriously. Instead of pushing doomerist messaging that shocks the public, there should be a visible, working effort to help Americans understand how alignment works and what risks might affect them in the near term. Today, this doomerist fervor galvanizes the public to pressure policymakers to act on an issue they have no technical understanding of. A majority of our nation's legislators do not understand AI and frequently spew talking points given to them by an intern, ChatGPT, or a lobbyist.

Ironically, Congress was better equipped to examine technological risks before 1995 than it is now. Congress created the Office of Technology Assessment to evaluate emerging technology, only to cut its funding in 1995. Its last director, Dr. Roger C. Herdman, who passed away in 2023, said in the final months after the office was defunded: "When congressmen look for advice and bring in the executive branch or the private sector, they will get the advice with the agendas and biases of those institutions."

Largely, the AI doomers have channeled the fear they manufactured around AI into support for whatever policy is presented as necessary for survival. If Americans come to accept that mass catastrophe is unavoidable without policies, then supporting them becomes an understandable reaction. This manufactured urgency forecloses the chance to examine whether a given proposal would actually work or how much control it would hand to companies or the government over people's lives. Now, to pass this as the public's consent, the small group of people in D.C. and San Francisco then points to this backing as approval for decisions made exclusively by them.

So, if the companies developing AI believe their work could put humanity in danger, or if the doomers inside those labs explicitly state that the technology they are building might end humanity, they should be expected—and even required—to explain exactly what they are doing and why they are continuing. Furthermore, if you are a doomer who believes AI might end humanity, your response should be to directly pressure these companies instead of routing that responsibility to everyday Americans.

The burden of proof lies with those accusing themselves of being a possible threat to the public. It does not lie with the American people, especially when the resulting policies concentrate power within a handful of frontier labs.

Anyone claiming to speak for the public must explain how their proposals answer that demand for control. If public anxiety becomes the justification for giving a handful of companies and officials more authority over everyone else's choices, Americans should actively question what is actually being accomplished in the name of the public or “saving humanity.”

Ultimately, it is consequential that Americans and our policymakers examine the technical and political assumptions behind any future presented as inevitable. We must understand what we gain or lose the right to do by accepting that version of the future. It is possible to choose a different technological future, and to build one in which progress gives people greater power over their lives and greater freedom to choose, imagine, and build. A claim about what might happen tomorrow cannot get to decide what Americans are allowed to do today.