Pushback clusters into four claims: self-regulation, the off switch, complexity, and China.
Before seatbelt legislation, the auto industry said safety mandates would destroy car sales. Before tobacco restrictions, the industry said government had no business regulating consumer choices. Before banning CFCs, the chemical industry said phase-outs were economically impossible. These were not fringe positions. The argument was advanced by credible people in credible publications, with credible-sounding reasoning..
The arguments against AI governance follow the same pattern. They sound reasonable on first hearing. They are often made by thoughtful people who believe them. They do not survive careful scrutiny.
What distinguishes AI governance from those earlier cases is the stakes. Leaded gasoline poisoned millions; it did not threaten civilization. A misaligned superintelligence may not offer a corrective period. Every objection to AI governance must be evaluated not just on its merits in isolation, but on the cost of accepting it if it turns out to be wrong.
Below are the four objections we encounter most often. Each is given its strongest form. Each is answered directly.
The companies building frontier AI have safety teams, publish safety research, and face reputational consequences for catastrophic failures. Voluntary commitments and market incentives are sufficient. Binding regulation would slow progress without improving safety.
The same argument was made about tobacco, leaded gasoline, financial derivatives, and social media. In each case, the industry's structural interest in growth was incompatible with adequate self-regulation. A company that voluntarily imposes strict safety requirements while competitors do not will lose market share. That dynamic does not change because the technology is more sophisticated. Self-regulation without binding rules is a lobbying position, not a safety strategy.
If an AI system becomes dangerous, humans can shut it down. The off switch ensures human oversight at every stage of development. The scenario where AI escapes human control is science fiction.
This assumes a sufficiently capable AI system will remain passive while humans deliberate about its future operation. A system advanced enough to pose existential risk will be capable of modeling the intentions of its operators and taking steps to preserve its continued operation if its goals require it. For a system operating at machine speed, the interval between "this may be a problem" and "this is an unrecoverable problem" may not allow a human response.
AI systems are too technically complex for legislators to govern effectively. Well-intentioned but technically uninformed regulation would cause serious harm without reducing risk. Leave it to the engineers who build the systems.
Governments regulate nuclear fission without legislators holding physics degrees. They govern pharmaceutical chemistry without requiring biochemistry PhDs. The skills required for AI governance are institution-building, treaty design, standards-setting, and accountability mechanisms, all things governments do constantly. The claim that AI is too technical to regulate is the argument that AI should be governed by the companies that build it. This is a commercial argument, not a technical one.
China is investing massively in artificial intelligence. The Chinese government has identified AI leadership as a national strategic priority and has committed hundreds of billions to achieving it. If the United States or Europe slows AI development through governance frameworks, China will reach artificial general intelligence first. A world where China achieves AGI before the democratic world is worse than a world where the West gets there first. Therefore, we cannot slow down.
This is the most coherent version of the China argument, and it deserves to be taken seriously. It is not a bad-faith position. Many thoughtful people hold it sincerely. The question is whether it survives scrutiny.
We believe it does not, for six distinct reasons. Each is addressed below. Taken together, they do not eliminate the competitive dimension of AI development, which is real. They do, however, demonstrate that the China argument, as typically deployed, provides no guidance and serves no strategic purpose beyond justifying the absence of any governance at all.
The alternative to international coordination is not a world where the democratic West wins the AI race and uses its lead responsibly. The alternative is a world where five unaccountable corporations build the most consequential technology in history with no oversight from any government, anywhere, under competitive pressure that ensures corners are cut. That is a vacuum pretending to be a strategy.
The core assumption embedded in the China argument is that China would benefit from being the first to build a misaligned superintelligence. A misaligned superintelligence is no weapon China can wield against the United States. It is an agent pursuing its own goals, which may be as incompatible with human survival in Beijing as in San Francisco.
China is also on this planet. Chinese leaders also have children and grandchildren. The scenario where China builds an unaligned superintelligence and uses it as a geopolitical tool against the West requires believing that Chinese leadership has solved the alignment problem that the best researchers in the world describe as unsolved. It also requires believing that Chinese leaders would risk the destruction of their own civilization for geopolitical advantage. The evidence for either belief is zero.
The United States had a nuclear monopoly in 1945. The Soviet Union developed nuclear weapons by 1949. Within four years, the world's two most powerful adversaries, with fundamentally incompatible political systems, active proxy wars, and genuine mutual distrust, had begun the process of nuclear arms control.
The Limited Test Ban Treaty came in 1963. SALT I in 1972. The INF Treaty in 1987. START in 1991. These agreements were negotiated with a state that US leadership called the "evil empire," verified by inspection regimes, and they worked. If verified arms control was achievable with the Soviet Union at the height of the Cold War, the argument that it cannot be achieved with China is simply not credible. The precedent exists. The techniques are known. Can the US and China agree on AI safety? →
China signed the Bletchley Declaration in November 2023, the first international agreement acknowledging that advanced AI poses potentially catastrophic risks. China participated in the Seoul AI Safety Summit in 2024. China has issued its own AI governance regulations, including requirements for algorithmic transparency, content labeling, and safety evaluations of frontier models before deployment.
The picture of China as a rogue actor unwilling to engage on AI governance is factually incorrect. China has strong reasons to want AI development to go well: its leadership is not immune to existential risk, its economy depends on technological stability, and a global AI catastrophe would destroy the prosperity that four decades of reform have built. The foundation for engagement exists. What is needed is the international framework that makes it concrete.
The logic of "we must race to stay ahead" assumes that being first produces the ability to control the outcome. But the opposite is true: the faster development proceeds, under competitive pressure with corners being cut, the less likely it is that the resulting system is safe, regardless of who built it.
A US lab that races to AGI under competitive pressure and without adequate safety evaluation does not produce a safe AGI that serves US interests. It produces an unsafe AGI that threatens everyone, including the United States. Competitive framing encourages exactly the development practices that make misalignment most likely. The race dynamic accelerates precisely the outcome everyone fears.
A persistent objection to AI treaties is that they cannot be verified. This claim is weaker than it appears. The hardware required to train frontier AI models passes through one of the most concentrated chokepoints of any technology in history.
TSMC fabricates virtually all cutting-edge AI chips. ASML makes the only lithography machines capable of producing those chips. NVIDIA designs the dominant GPU architecture. All three companies are in Allied countries or under Allied export control jurisdiction. The United States government already attempts to restrict the flow of advanced AI chips to China through export controls. A treaty-based compute monitoring regime would formalize and extend what is already being done unilaterally. Verification is an engineering problem. Engineering problems have solutions.
The deepest problem with the China argument is its real-world function, not its accuracy about China. The argument is used to justify doing nothing at all: no domestic regulation, no international engagement, no safety requirements, no liability frameworks, no governance of any kind.
If the answer to every AI safety proposal is "but China," then no action is ever possible. This is a veto dressed in geopolitical language. It serves the interests of the companies racing to build AGI, who benefit from a political environment so paralyzed by competitive anxiety that no binding rules can ever be enacted. The China argument is real. The way it is deployed is a strategy for permanent inaction. We should not accept it as such.
Some of the loudest lines against governing superintelligence are not careful forecasts. They are psychological moves. Open any item for the move, and the counter.
This is learned helplessness dressed as realism. If the outcome is fixed, effort is pointless, so you stop organizing. Historians call a close cousin of this "false finality": treat a live political choice as already settled so the audience never tries. Nuclear weapons were also called inevitable for every state that could afford them. Most never got them. The Montreal Protocol was called impossible until it wasn't.
A technology is inevitable only if people keep choosing to build it and refuse to bind one another. That is a choice, not a law of physics. The people who say "inevitable" the loudest often stand to gain from your passivity.
This converts a coordination problem into a surrender speech. It assumes the rival is a cartoon who will never join a verified halt, and that your only options are race or quit. Adversaries with worse blood between them still froze whole classes of weapons when mutual suicide was the alternative. Stopping CFCs did not end refrigeration. Stopping superintelligence does not end useful AI.
Label the person so you never have to answer the claim. Call someone a doomer and the room is invited to dismiss the argument as a personality type. The same move has been used against seatbelts, tobacco limits, and climate science. Name-calling is not a safety case. "Doomer" is not an argument →
Appeal to authority without the authority. Many of the people closest to the work, including Geoffrey Hinton and Yoshua Bengio, have said the risk is real and under-managed. Internal safety teams sit inside firms whose incentives reward shipping. "Trust us" is not verification. Verification is inspectors, thresholds, and costs for cheating.
Another form of conditioned passivity. If the window is already closed, you can skip the hard work of politics. Capability is moving fast. Treaty work is still open, and frontier training still runs through a short list of chipmakers and fabs. Late is not the same as over. Acting like it is over is how it becomes over.
The United States and the Soviet Union competed intensely on every dimension of power during the Cold War and simultaneously built an arms control architecture that prevented nuclear annihilation. The two things coexisted because both sides recognized that some outcomes were worse for everyone, regardless of who caused them.
The same logic applies to AI. The United States and China can compete on AI applications, on economic deployment, on research and development, while simultaneously agreeing that certain categories of AI development (specifically systems whose capabilities exceed human intelligence across all domains and whose alignment cannot be verified) should not be built by either side without multilateral verification frameworks in place.
This is exactly the position that was taken, successfully, on nuclear weapons. The people who negotiated the NPT and the SALT treaties were no naive idealists about Soviet intentions. They were hard-headed realists who understood that mutual annihilation was not in the interest of either side, regardless of ideology.
We advocate for the same hard-headed realism on AI. China and the United States both face the same existential risk from misaligned superintelligence. A treaty that prevents both from building it without adequate safety verification serves the interests of both. The negotiation will be difficult. The alternative (a race to build the most consequential and potentially dangerous technology in history with no governance and no safety verification) is worse for everyone, including the parties who say they want to win.
The task is to pursue international coordination with the same determination that produced the nuclear non-proliferation treaties, the chemical weapons conventions, and the ozone agreements. It was done with adversaries before. It can be done again.
Читайте наш полный планJoin those who are building the case for governance before it is too late.