A report published on March 2 by The Wall Street Journal, titled "Incompatibility Leads to Pentagon Parting Ways with Anthropic," is translated below:
In his first face-to-face meeting with U.S. Defense Secretary Peter Hegseth, Anthropic CEO Dario Amodei outlined his views on the risks of artificial intelligence (AI) controlling autonomous weapons.
Hegsays had no interest in hearing this, even though it came from a CEO whose company's AI tools had become a valuable asset to the U.S. military.
According to sources, during the meeting on February 24, Hegsays interrupted Amodei and said, "No CEO of any company should try to tell our combat personnel what to do."
Today, the Trump administration, which has been advocating for the rapid rollout of AI and viewing it as critical to economic growth and national security, finds itself locked in a stalemate with a leading domestic AI company.
Michael Horowitz, a former U.S. Department of Defense official who was involved in AI policy-making, said: "It's essentially a battle of charisma and personality, just disguised as policy differences."
He said the conflict ultimately stemmed from "a breakdown in trust between Anthropic and the Pentagon, because Anthropic didn't trust the Pentagon enough to know how to use its technology responsibly, and the Pentagon didn't trust Anthropic to be willing to work for the important applications it needed."
On February 27, US President Trump instructed all federal agencies to cease cooperation with Anthropic and attacked the company's executives as "left-wing psychopaths."
Later that day, Hegsays designated Anthropic as a supply chain risk company because the company failed to agree on an agreement regarding the use of its tools by the deadline.
Ironically, according to The Wall Street Journal, just minutes before Trump posted the message, he authorized strikes against Iran, and the planning of these strikes used Anthropic's "Claude" model.
For years, Anthropic has been one of the most vocal advocates for safeguards to ensure the safe application of its technology within the AI sector. This stance has sometimes frustrated government officials who have widely deployed Anthropic's tools across government departments but are also troubled by the company's attempts to control how those tools are used.
Last week, U.S. Under Secretary of Defense for Research and Engineering Emil Michael accused Amodei of trying to play God. A government official stated that other tech company CEOs don't tell the government how to use their technology, but instead seek compromises.
Critics see these moves as the latest example of the current government forcing a private company to comply.
At the heart of this conflict lies a new question: in war and in broader social life, who should ultimately control how cutting-edge AI tools are deployed?
What will happen next remains to be seen, but this standoff appears to be boosting Anthropic's popularity among consumers. As of March 1, the "Claude" model had surpassed OpenAI's ChatGPT to become the most downloaded app on Apple's App Store.
This article is a user submission and does not represent the views of this website.
The copyright of this content belongs to the original author. Please contact the original author for authorization before reprinting. For any copyright infringement issues, please contact copyright@jaketao.com
