The US military is integrating artificial intelligence into its operations. According to informed sources, OpenAI is collaborating with two defense technology companies selected by the Pentagon to develop voice-controlled drone swarm combat software for a US military competition. The technology from OpenAI is being used to convert battlefield commanders' voice commands into digital instructions executable by drones.
Separately, The Wall Street Journal reported that the US military used Anthropic's AI tool, Claude, during a January operation targeting the capture of former Venezuelan President Nicolás Maduro. This highlights the increasingly significant role AI models are playing for the Pentagon.
OpenAI's involvement is part of a $100 million military challenge initiated in January by the Defense Innovation Unit and the Special Operations Command's Defense Autonomous Operations Team. The competition seeks prototypes capable of commanding autonomous drone swarms via verbal orders. OpenAI's role is limited to translating spoken battlefield commands into digital instructions for unmanned systems; its technology does not control drones, integrate weapons, or have targeting authority. The company stated it did not submit its own bid and its participation is limited to providing an open-source version of its model.
The six-month competition will proceed in phases, starting with software development and moving to field testing. Later phases envision multi-domain coordination across air and sea systems, with the Pentagon noting that mission execution elements could impact system lethality and effectiveness. This initiative marks a broader expansion of OpenAI's defense collaborations. This week, the Pentagon also announced a separate arrangement to provide ChatGPT to approximately three million Department of Defense personnel. CEO Sam Altman previously stated the company does not anticipate developing AI-powered weapon platforms in the foreseeable future but has not ruled out the possibility.
Regarding the use of Claude, sources indicated the US Department of Defense accessed the tool through Anthropic's partnership with data analysis software firm Palantir, a frequent collaborator with the Defense Department and federal law enforcement. Anthropic is the first AI model developer whose technology has been used by the US Defense Department for classified operations. The operation in Venezuela may have involved other AI tools for unclassified tasks, with potential uses ranging from document summarization to controlling autonomous drones.
While military adoption is seen as a significant boost for AI companies seeking legitimacy and supporting high valuations, Anthropic executives, including CEO Dario Amodei, are highly concerned about the potential societal risks of AI. An Anthropic spokesperson declined to comment on whether Claude or any AI model was used in any specific operation, stating that all use of Claude, whether by private sector or government agencies, must comply with the company's usage policies governing deployment. The US Department of Defense declined to comment.
In a January event announcing Pentagon collaboration with xAI, Defense Secretary highlighted that the military would not employ AI models that prohibit their use in warfare, remarks perceived as a reference to discussions with Anthropic. In January, the US Department of Defense released a new AI strategy and related memorandum aimed at building an "AI-first" combat force to consolidate US military advantage. The strategy explicitly targets accelerating the establishment of AI dominance within the US military, with its deployment approach and implementation pathways warranting attention.