The US has decided to shoot itself in the foot and destroy one of the best AI companies in the country

Dario Amodei, CEO of Anthropic, published a few hours ago a statement in which he announced something unusual: the Department of Defense (DoD) has confirmed that “we have been designated as a risk to the national security supply chain” of the United States. This agency thus fulfills the threat it posed a few days ago and automatically turns Anthropic, one of the best AI companies in the country (if not the best) into a pariah company. What implications does that have? Many and all of them huge.

I veto Anthropic. This designation prohibits Anthropic from doing business or developing projects for the US military. That is already serious, but it is not just the Pentagon, for example, that will not do it: any company that works with the Pentagon is also prohibited from using Anthropic’s AI services for any government project. We are facing a decision whose collateral effects could be terrible for Anthropic. The loss of revenue could be massive, and if other federal agencies follow the Pentagon’s lead, Anthropic could have a hard time defending its viability against its competitors. That designation is not immediate, and there will be a transition period six months for DoD to migrate to other vendors (like OpenAI).

It had never been done with a national company. The ban on Anthropic is absolutely extraordinary, and that designation as a “supply chain risk” was a measure historically reserved for foreign adversaries like Huawei. By applying this label to an American company, the DoD severs its commercial ties and marks the company with a stigma, a kind of “scarlet letter” that could scare away global investors and partners.

ethical shock. The core of the conflict is not technical, but moral. Anthropic was born as a spin-off from OpenAI with the aim of avoid existential risks in the development of AI models, and the company has always positioned itself as a great defender of alignment with human values. Its CEO, Dario Amodei, insisted that its AI could not be used for mass surveillance or for the development of lethal autonomous weaponsbut that has collided head-on with the US government and military establishments, which wanted practically total access without restrictions, except those imposed by the US Constitution and laws.

to the courts. Amodei has explained in its statement that it will fight the decision in court. His argument, he explains, is that statute 10 USC 3252 It is a tool of protection, not punishment. The defense will need to focus on showing that the Department of Defense did not use the least restrictive means to ensure security. If they succeed, they could invalidate the designation, although the reputational damage has already been done.

The dilemma of sovereignty. Can a private company be above the Government? The Pentagon argues that no supplier can slip through the chain of command, and one thing is certain here: for an AI to have usage clauses that limit military operations is to cede national sovereignty to a private algorithm and the terms of service of a board of directors and a CEO who have not been democratically elected.

The threat of extreme interventionism. This unusual measure could end up setting a precedent. If the government punishes companies that ask uncomfortable questions or place limits on the use of their technology, AI innovation could change its philosophy. Companies that want to survive would have to do so without questioning the orders out of pure fear of bankruptcy and bankruptcy.

Transition period. There is, however, a period of six months granted for the transition and that seems to make it clear that the Pentagon still depends on Anthropic technology for current operations, as demonstrated by the kidnapping of Nicolás Maduro or the current intelligence analysis of the conflict in Iran. It remains to be seen how events will evolve, but the outlook for Anthropic is certainly worrying. And for the rest of the companies too, if indeed justice rules in favor of the Department of Defense.

Image | Anthropic | Xataka with Freepik

In Xataka | Anthropic has become the Apple of our era and OpenAI our Microsoft: a story of love and hate