In America, Even a Single Company Can Challenge the Government

U.S. President Donald Trump declared a complete government ban on Anthropic on April 27, just before launching airstrikes on Iran, calling the company a radical left-wing "woke" enterprise.

Yet barely half a day after he said "America will never allow Anthropic to dictate our military," the Pentagon revealed at 10:45 a.m. on April 28 that it had used Anthropic's AI tool Claude to execute the Iran strike. The company had proven its capabilities in a way that made the words of the "hegemonic" president ring hollow.

While Elon Musk's xAI and Sam Altman's OpenAI volunteered to replace Anthropic, the Defense Department chose to stick with the company through the remaining contract period. Exactly why Anthropic proves overwhelmingly superior for warfare compared to competitors remains unclear. What is known is that the company has learned from vast amounts of classified Pentagon data and helps make attack decisions on actual battlefields.

Anthropic's general characteristics—training on human-written books rather than data scraped from the internet, self-coding capabilities designed for developers from the start, and compatibility with legacy systems—likely also benefited the Pentagon. However, with competitors announcing new AI updates almost daily, Anthropic could soon lose its current position.

What draws attention to Anthropic, led by Dario Amodei, is not its technology but the ethics he espouses. The sight of a 44-year-old CEO of a "mere" private company boldly pushing back against White House measures on ethical grounds came as a refreshing shock even in America.

In a CBS interview shortly after the Pentagon's termination notice, Amodei carefully but clearly stated his convictions. He emphasized that 99% of applications could proceed as the Pentagon wished—excluding only two things: mass surveillance of Americans and fully autonomous lethal weapons.

He argued that social consensus through Congress must come first before robots attack in place of humans. While immediate warfare matters, he pointed to the risk of robots mistakenly killing people and the dangerous concentration of power when one person controls 10 million drones. His remarks poured cold water on Wall Street's Anthropic fever while forcing AI enthusiasts to reconsider their positions.

It was also unusual for a company at the forefront of technological development to suggest matching Congress's slower pace. While companies have long complained that Congress holds them back, Amodei believes that precisely because companies witness rapid technological change, they must alert lawmakers to emerging problems. His view that technology can evolve only after all members of society reach consensus—not just the smartest developers and most powerful leaders—is remarkable.

Of course, Amodei's ethics have weaknesses, and insiders have come forward with allegations. His principles may drift as Anthropic's valuation grows too large for ethical emphasis and investor pressures mount. His public disclosure of negotiation details could ultimately be seen as a bargaining tactic against the Pentagon. The realist argument—what if adversaries develop AI-powered autonomous weapons in actual warfare?—is difficult to refute.

Still, Amodei's Anthropic has sounded an alarm for the global AI industry racing forward at full speed. What is more striking is that the U.S. government continued to choose Anthropic. Even after the company publicly criticized the government, disclosing negotiation details and arguing there was no legal basis, Washington ignored the criticism and partnered with Anthropic out of necessity.

This stands in stark contrast to Korea's industrial ecosystem, where companies fall in and out of favor with each change of administration, causing business fortunes to fluctuate accordingly. It appears another reason has emerged for why America's AI industry can advance further.