A federal court decision known this Friday provided new insights into the use of artificial intelligence by the United States during the military operation that culminated on January 3 with the capture of Nicolás Maduro in Venezuela, although the document does not disclose the specific role that the technology played during the mission.
The journalist David Alandete, correspondent at the White House, reported this Friday on X that the ruling confirms the presence of Claude, the artificial intelligence model developed by Anthropic, within the technological framework used by the Pentagon through Palantir.
The ruling from the U.S. Court of Appeals for the District of Columbia Circuit, dated September 25, was issued in the litigation of Anthropic against the Pentagon's decision to exclude the company from its supply chain due to its restrictions on certain military uses of Claude.
The ruling states that, in early 2026, An executive from Anthropic questioned the use of Claude by a contractor in a "sensitive military operation abroad."
That intervention raised concerns within the Department of Defense about whether the limitations incorporated into the model could affect its operation during an ongoing mission.
The court clarifies that the Pentagon did not explicitly identify in the record what that operation was.
However, Anthropic included in the case news reports indicating that the company had raised concerns with Palantir about the role of its technology in the military operation on January 3 to capture Maduro.
The ruling later revisits the episode and notes that it is not exactly known what happened during that incident, but it mentions the possibility suggested by Anthropic that it could have occurred during a “kinetic operation” aimed at capturing a foreign head of state.
In February, Reuters reported on a piece from The Wall Street Journal stating that Claude had been used during the U.S. operation in Venezuela through Anthropic's relationship with Palantir Technologies, whose platforms are widely utilized by the Pentagon and federal agencies.
The ruling now adds an additional element: it confirms that the controversy between Anthropic, Palantir, and the Pentagon regarding the operation of Claude during a foreign military operation was formally documented in the court record, and that the reports on Maduro's mission were submitted by the company itself in the proceedings.
Claude was also not an unfamiliar tool to the U.S. national security community. According to the ruling, since 2024, the Department of Defense and intelligence agencies had been using commercial versions of the model in classified systems through contractors.
Among the tasks mentioned in the file are processing classified documents, summarizing threat assessments, and translating intercepted materials.
Subsequently, Anthropic developed specific versions of Claude for national security agencies.
The company explains that its Claude Gov models were created specifically for U.S. national security customers and can be used in strategic planning, operational support, intelligence analysis, and threat assessment within classified environments.
Anthropic has also publicly acknowledged that Claude has been employed within the U.S. security apparatus for intelligence analysis, modeling and simulation, operational planning, and cyber operations.
Alandete noted that, in the case of Venezuela, among the possible functions of a tool with these characteristics could have been the processing of large amounts of intelligence, the analysis of surveillance information, tracking of individuals or vehicles, and support for operational planning.
The dispute arose precisely over the limits that Anthropic imposed on the military use of its models. The file shows that the company gradually expanded the permitted uses for the Government, including foreign intelligence analysis, weapon systems design, and offensive cyber operations, but maintained two restrictions: the use of Claude for fully autonomous lethal weapons and for mass surveillance of Americans.
The Pentagon later demanded that its suppliers allow the use of models for "any legal use". Anthropic refused to remove those two restrictions, and Secretary Pete Hegseth ordered in March to exclude the company's products from military systems and prohibit contractors from using them for Department work.
This Friday, the appeals court upheld the Pentagon's decision by concluding that there were sufficient grounds to consider that the restrictions on Anthropic could pose a risk to national security operations if Claude stopped responding as expected during a mission. The decision was made by a vote of two to one.
After the conflict, the Department of Defense also began to expand its contractual relationship with OpenAI, according to the court record itself.
The revelation adds a new technological dimension to the U.S. operation that culminated in the capture of Maduro and his transfer to New York on January 3rd.
Days after his arrival in the United States, Maduro appeared before federal court and pleaded not guilty to the charges brought against him. Since then, he has been held in U.S. custody while the judicial process continues.
Related videos:
Filed under: