Dario Amodei may be one of the most consequential businessmen Americans have never elected to anything. As CEO of Anthropic, the company behind Claude, Amodei sits near the center of the race to develop increasingly powerful artificial intelligence. He also possesses unusually strong views about how that technology should be controlled, which countries should possess it, what governments should be permitted to do with it, and how AI might reshape society.
That raises an uncomfortable question: Are Amodei’s motives nefarious?
There is little publicly available evidence supporting the dramatic interpretation that Amodei is secretly pursuing some malicious agenda. His writings suggest something considerably more complicated—and perhaps more important. He appears genuinely convinced that extraordinarily powerful AI is approaching rapidly, that it could produce enormous benefits, and that it simultaneously poses potentially catastrophic dangers. In his 2024 essay “Machines of Loving Grace,” Amodei described AI as potentially capable of accelerating advances in medicine, neuroscience, economic development and poverty reduction.
His more recent writings have emphasized the darker side of that equation. Amodei worries about autonomous AI systems, biological weapons, economic disruption, authoritarian governments and concentrations of technological power. Importantly, he has even argued that AI companies themselves require scrutiny because conventional corporate governance may prove inadequate for organizations controlling enormously powerful systems.
That hardly sounds like a cartoon villain.
But conservatives should understand that benevolent intentions are not the same thing as benign consequences.
The most revealing element of Amodei’s worldview may be his conviction that those developing advanced AI have responsibilities extending well beyond producing useful technology. He has openly contemplated how AI could strengthen liberal democracy, improve government services, influence international power and help democratic countries maintain superiority over authoritarian rivals. He has proposed an “entente strategy” under which democratic nations maintain an AI advantage, restrict adversaries’ access to critical technologies and use that advantage to strengthen a democratic international coalition.
One can agree with much of that objective while still being troubled by the premise.
The conservative concern should not necessarily be that Amodei wants tyranny. It is that Silicon Valley’s most powerful figures increasingly believe technological expertise grants them a special role in determining the boundaries within which society should operate.
That problem became particularly visible during Anthropic’s 2026 confrontation with the federal government. Amodei insisted that Anthropic strongly supports using AI for American national defense and noted that its technology was already being used for intelligence analysis, cyber operations, operational planning and other national-security functions. Yet Anthropic refused to remove restrictions involving mass domestic surveillance and fully autonomous weapons.
There is considerable merit in those concerns. Conservatives traditionally distrust warrantless government surveillance and should hardly welcome machines independently deciding when Americans—or anyone else—should be killed.
But there is another constitutional principle involved. Anthropic is a private corporation, not an elected branch of government. Its executives should not become a substitute legislature deciding America’s national-security doctrine merely because their technology has become indispensable.
This is where the question of Amodei’s motives becomes less important than the question of Amodei’s power.
Perhaps he is entirely sincere. Perhaps he genuinely believes he is protecting humanity. His public record certainly provides substantial evidence for that interpretation. Anthropic has simultaneously supported American AI leadership, worked with national-security agencies, advocated restrictions on advanced technology reaching authoritarian competitors, and argued for safeguards against government misuse.
Yet history offers plenty of warnings about intelligent people becoming convinced that their superior understanding justifies extraordinary influence over everyone else’s choices.
The danger does not require conspiracy.
It requires only concentrated power combined with certainty.
Artificial intelligence companies are accumulating something unprecedented: technological capabilities that governments increasingly need but may not completely understand. That creates the possibility of an inversion of traditional democratic authority. Instead of elected governments establishing policy and companies supplying technology within those boundaries, technology companies could increasingly establish the permissible boundaries and force governments to operate inside them.
The reverse is equally dangerous. Government should not be allowed to commandeer private AI companies and demand unlimited access to their technology merely by invoking national security.
The conservative answer, therefore, should not be deciding whether Dario Amodei is hero or villain. It should be refusing to construct a system that requires Americans to trust him enough for the distinction to matter.
Amodei himself has written that AI companies should be carefully watched and that their relationships with government require limits. That is one principle on which conservatives, progressives and technologists ought to find considerable agreement.
The people building extraordinarily powerful machines should not govern America. Neither should bureaucrats controlling those machines.
Congress must establish clear constitutional and statutory boundaries while meaningful democratic oversight remains possible.
Dario Amodei’s motives may ultimately be honorable. But the American system was deliberately designed around a more durable principle: Free societies should never depend upon the benevolence of powerful men.