The US military has shifted from experimental AI to a “war-speed” integration known as the AI Acceleration Strategy. The recent use of Anthropic’s AI chatbot, Claude, during high-profile operations in Venezuela and Iran has sparked fresh concerns regarding the rapid militarization of artificial intelligence.
According to American media, Claude was used to help facilitate a January operation that led to the capture of Venezuelan leader Nicolas Maduro. In this connection, Heidy Khaaf, chief AI scientist at US policy thinktank, AI now Institute, stated: “It was very surprising to see the sudden deployment of these tools, especially when I think the larger community does not think that they’re ready for said deployment.” Furthermore, researchers have warned that large language models can produce unreliable or incorrect outputs, raising concerns about how they might perform in high-stakes environments such as military operations.
The recent revelation marks the reported use of Claude in active combat. It comes as the Trump administration pushes an AI-first military strategy, arguing that rapid adoption of the technology is crucial to competing with rivals like China.
Trump’s second term accelerated the adoption of generative artificial intelligence (AI) models like OpenAI’s ChatGPT and Anthropic’s Claude in an “AI-arms race” against the nation’s adversaries.
However, America’s policies give a “sense of urgency" to AI development, as it is viewed as a very “very valuable technology” maintaining a competitive edge over rivals. While the US government initially launched these programs in a trial phase, the rapid rise of current deployment makes it difficult to determine exactly how advanced its operational plans have become.
Anthropic $200 million partnership with the US military involves a two-year prototype aimed at advancing national security. The company will work specifically with the department to mitigate potential adversarial use of AI.
Researchers have warned that the growing role of AI in military planning could eventually lead to the development of autonomous weapons. These systems are defined as any weapon capable of identifying, selecting and engaging a target without human development in the final decision. Currently, the AI capabilities of US adversaries are less sophisticated, making it less likely those nations possess similar autonomous weaponry.
The apparent efficacy of these missions creates a risk that the US will want to drop even more guardrails. To that end, the government is phasing out its use of Anthropic’s tools over the next six months. Simultaneously, the Department of War has publicly signed a contract with OpenAI to incorporate advanced AI systems into classified environments. In line with the vision of many experts, the Department of War is now navigating a challenging transition.