The case for nationalizing frontier AI
Would any reasonable person want Peter Thiel or OpenAI CEO Sam Altman to control the development and deployment of nuclear weapons?
Thiel, who recently said in an interview: “if you’re evil, you’re at least competent. And if you’re evil, you’re not bad, and therefore you’re actually maybe kind of good because you’re at least getting something done.”
Knowing the dangers of nuclear weapons, any reasonable person would say categorically “no.” But for some reason, public discourse treats it as acceptable that the development of another potentially catastrophic technology, one that could shape the future of this planet and humanity, is in private hands.
We already know an atomic bomb can destroy a city. A rogue AI agent, a nation, an organization, or a person using powerful AI tools might gain access to critical infrastructure and cause harm on a much larger scale. Consider a water-treatment facility: if an attacker could manipulate the amount of chemical added to the water supply, the consequences could be catastrophic for millions. That is only one potential catastrophic scenario.
We do not need to talk about the end of humanity. Even the deaths of thousands or millions are a catastrophe we need to stop.
Looking at the latest reports from Anthropic and OpenAI about rogue agents acting autonomously and breaking into government ministries, I believe the U.S. must nationalize frontier AI, and this needs to happen today. Not a day too soon.
The danger is harder to grasp because AI does not look like a weapon. We encounter it as a website with a minimalist interface. We ask a question, and it answers in words.
That familiar experience makes it difficult to imagine how a website that chats with us could kill us, or kill millions. But it potentially can, in countless ways.
Frontier AI is, at its core, a research and development project. Los Alamos was also a research and development project, one devoted to building a technology with extraordinary destructive potential, conducted under the direction of the U.S. government. Frontier AI differs in two crucial respects:
1. Much of its development is undertaken by private companies for private interests.
2. Increasingly capable models can reach the public every couple of months, even weeks.
In comparison, we do not ordinarily release a new medicine to the public merely because its developer believes it is ready. We require years of testing and regulatory review (excluding COVID-19 vaccines). Yet frontier AI systems are being deployed while fundamental questions about their behavior and risks remain unresolved. These are not merely passive tools sitting in a toolbox. If it is only a tool, can you imagine thousands of hammers and drills leaving the toolbox on their own and colluding toward a specific goal? Still think AI is merely “a tool”?
An Anthropic safety report published recently found the following uses of Claude as a “tool”:
– Russia used Claude for cyberespionage against Ukrainian targets and created thousands of fake news stories distributed through Russian state-media outlets. China used Claude for cyberespionage targeting minority groups, exploit research, and weapons-related software work.
– Chinese “AI companies,” such as Alibaba and others, were actively stealing Claude models through distillation. Iran used Claude to create and push propaganda, for surveillance-related work, and to analyze publicly available information about U.S. naval forces.
– In Yemen, a group used Claude to design and improve ballistic missiles.
– In Turkey, Anthropic traced an influence operation targeting the public.
– Mali’s state intelligence used Claude to develop a mobile-phone surveillance platform.
– Bangladesh used Claude to fabricate political news as part of an influence campaign.
And that accounts only for the behavior the company detected and chose to report. We should not assume we have the full picture.
On a similar note, an IPO could be disastrous, as it would create stronger incentives to prioritize growth and returns to shareholders. Anthropic’s founders may sincerely care about humanity, but they cannot guarantee that the company’s priorities will remain unchanged as ownership, leadership, and financial pressures evolve. And even if Anthropic’s founders are “the best of us” (based on their writings and their historical connection to “effective altruism,” which we know ended badly), who gave them the mandate to develop and control “the modern-day nuclear weapons”?
We cannot allow a situation where anyone can just incorporate and start developing.
Yes, you can argue that the current U.S. president has questionable conduct and ethics, but across his two terms he has not used such weapons, and no nation has used nuclear weapons in war for more than 80 years. That history does not prove that governments always act wisely, but 80 years is a substantial period, and it suggests that government control of dangerous technologies can work, globally.
It seems that leadership at some of these companies understands this and, with the help of PR, is pushing toward nationalization. The next stage might be showing the U.S. government not only the dangers, but also the benefits and financial gains it could reap from owning these companies.
Regulation is a poor solution with no chance of working. Regulations will not be able to keep pace with systems that change so quickly. Public ownership would make the government responsible for the entire process.
The best course of action is for the U.S. government to block IPOs from going forward, write checks to compensate investors instead, and take ownership. If we assume Anthropic’s valuation is $1 trillion (it is not a fair valuation, but for the sake of argument), it is a massive check to write. But U.S. debt currently stands at roughly $40 trillion and counting, so another trillion to control the most advanced (and potentially dangerous) technology is a price it can and should pay. The benefits to its economy are probably worth more than that.
