Anthropic's AI played a key role in U.S. strikes on Iran - The Washington Post

The U.S. military actively used a platform in which artificial intelligence from Anthropic - Claude was integrated in strikes against Iran, The Washington Post reported, citing sources familiar with the situation. The analytics platform in question helped the military analyze intelligence and rank targets in real time, the sources told the newspaper. Also, according to them, the U.S. used this platform "to counter terrorist plots" during the capture of Venezuelan President Nicolas Maduro.

Claude is integrated into the Maven Smart System, a military system developed by Palantir Technologies and used by the U.S. military. According to two of The Washington Post's sources, over the past year Claude combined with Maven has evolved into a tool that is now used on a daily basis in most U.S. military units. According to the Washington Post, the AI-integrated platform, despite a dispute between Claude's developer, Antropic, and the Pentagon, has been used by the U.S. military, among others, in strikes against Iran to hit about 1,000 targets in the first 24 hours of the operation.

During the planning phase of a strike against Iran, the Claude-based Maven system suggested hundreds of targets, provided precise coordinates of their locations and prioritized them according to their importance, two sources told the publication. The combination of Maven and Claude has created a tool that accelerates the military campaign, reduces Iran's ability to retaliate and turns operations planning, which used to take weeks, into a near real-time process, one of the sources told the newspaper. Artificial intelligence tools also assess the results of strikes after they have been launched, he said.

Claude, according to the publication's interlocutors, was also used by the U.S. military "to prevent terrorist plots" during the operation to capture Venezuelan President Nicolas Maduro. This was the first time an AI system was used in a major military operation, the sources told The Washington Post.

According to the Palantir website, Maven was designed, among other things, "to improve interoperability between strategic and tactical operations of different military departments." Among others, the platform has been used by the U.S. Army, Air Force, Space Force and U.S. Navy since 2024. As of May last year, the system was used by more than 20,000 servicemen, The Washington Post reports.

The platform, the newspaper explains, analyzes large amounts of classified data from satellites, surveillance systems and other intelligence sources to help identify and prioritize targets in real time.

A few hours before the bombing of Iran, US President Donald Trump announced a ban on the use of Anthropic tools by US government agencies - he gave government agencies six months to completely abandon these technologies. The statement was made against the background of pressure on AI developer Claude from the U.S. Department of Defense: the department demanded that the company, with which he has a contract, give him full access to AI functions. Anthropic, in turn, sought guarantees from the Pentagon that its AI models would not be used for fully autonomous weapons or for mass domestic surveillance of Americans.

The $200 million contract between the AI startup and the U.S. Department of Defense was signed last July. It included, among other things, work with classified materials, CNBC recalls.

According to public statements, the AI system into which Claude is integrated is used by the military to generate target suggestions, track logistics and produce intelligence summaries from the field, The Washington Post reported.

The Pentagon did not respond to journalists' request about Maven's role in the military operation against Iran. Representatives of Anthropic and Palantir declined to comment. The Wall Street Journal also reported on the use of Claude in strikes against Iran on February 28.

The military will continue to use the Maven platform with Claude integrated into it until a replacement is implemented, two sources told The Washington Post. Other AI companies are already poised to replace Anthropic at the Pentagon, the publication notes. Elon Musk's xAI, for example, as well as Anthropic's main competitor OpenAI, signed agreements last week to work with classified government systems.

This article was AI-translated and verified by a human editor