Palantir CEO Clarifies AI Use with the Pentagon
In a recent interview with Fortune, Alex Karp, the CEO of Palantir, addressed concerns regarding the potential use of artificial intelligence (AI) by the U.S. Defense Department. Karp emphasized that the Pentagon is not utilizing AI for domestic surveillance of American citizens and has no plans to do so in the future.
Palantir, a major player in defense software based in Miami, provides crucial data analytics and AI tools to the Department of Defense, commonly known as the Pentagon. Their software is the primary method through which the Pentagon employs Claude, Anthropic’s large language model.
During a conference held by Palantir, Karp reiterated, “We are still in the midst of this development.” He further explained that Palantir is responsible for running these AI models. Following a partnership established in 2024, Anthropic has been directly collaborating with the Defense Department to tailor its technology for military use.
Karp took a stance favoring the Pentagon, stating that there was never an intention for these tools to be used domestically. He framed the conversation as focused on non-American citizens in a combat setting.
The Clash Between Anthropic and the Pentagon
For over a year, Anthropic’s AI model, Claude, has been the preferred tool for the U.S. government and was recently cleared for classified operations. Reports suggest that Claude played a role in capturing Venezuelan President Nicolás Maduro. However, tensions between Anthropic and the Pentagon have escalated recently, leading to a significant announcement from the Trump Administration, designating Anthropic as a supply-chain risk to national security for the first time.
The conflict revolves around the safeguards that Anthropic requested, aiming to ensure Claude would not be used for domestic surveillance or in lethal autonomous weapons. The Pentagon insisted that Claude be accessible for all lawful applications, but Anthropic raised concerns.
In a blog post, Anthropic outlined two key reasons for its position. First, they contend that current AI models are not sufficiently reliable for fully autonomous weapons, as this could jeopardize the safety of both military personnel and civilians. Second, they maintain that mass surveillance of Americans undermines fundamental rights.
Apologies and Legal Action
Following the fallout, Anthropic issued an apology for a leaked internal memo that criticized OpenAI and its CEO for accepting a deal with the Pentagon shortly after Anthropic opted out. The company clarified that this memo was not intentionally leaked and did not reflect their current views.
However, shortly after the apology, Anthropic decided to take legal action against the U.S. government for designating it as a national security risk. They filed a lawsuit in the U.S. District Court for the Northern District of California, claiming that the Pentagon’s actions were “unprecedented and unlawful.” The complaint argued that the government’s decision infringed on the company’s rights and highlighted that no federal statute permits such actions.
In a complex situation involving AI technology and government regulations, the outcome of this legal battle could have significant implications for both companies and the future of AI in defense applications.
