收听本期播客
阅读正文
A recent report by the Wall Street Journal has brought to light a significant and controversial development concerning the integration of artificial intelligence into military operations. The report suggests that the US military reportedly deployed ‘Claude’, an AI model developed by the company Anthropic, during an operation in Venezuela. This incident is presented as a high-profile example of the US Department of Defense’s increasing reliance on advanced AI technology.
According to the Wall Street Journal, the objective of this covert mission was to kidnap Nicolás Maduro, Venezuela’s president. Venezuela’s defence ministry reported that the operation involved bombings across the capital, Caracas, tragically leading to the deaths of 83 people. This alleged use of Claude raises serious ethical and policy questions, particularly given Anthropic’s own terms of use. These policies explicitly forbid the deployment of Claude for violent purposes, for developing weapons, or for conducting surveillance, which appears to directly contradict the nature of the reported military action.
Neither Anthropic nor the US Department of Defense has officially commented on the claims regarding Claude’s involvement. A spokesperson for Anthropic did issue a statement, clarifying that any application of their AI tools must strictly adhere to its established usage policies. The Wall Street Journal cited anonymous sources who indicated that Claude was utilised through Anthropic’s partnership with Palantir Technologies, a major contractor for the US defence department. Palantir also chose not to comment on the allegations. It remains unclear precisely how the AI tool, which possesses diverse capabilities ranging from processing documents to potentially piloting drones, was employed during the operation.
This event highlights a global trend in modern warfare, where armed forces worldwide are increasingly incorporating AI into their strategies and arsenals. For instance, Israel’s military has used autonomous drones in Gaza and extensively relied on AI for targeting. Similarly, the US military has employed AI for recent strikes in Iraq and Syria. However, this growing dependence on AI is met with strong warnings from critics. They express significant concerns about the deployment of autonomous weapons systems, pointing out the inherent risk of targeting errors and the complex ethical dilemmas that arise when computer algorithms are tasked with determining who should or should not be harmed.
The situation also exposes a growing tension between AI developers and defence agencies. Anthropic’s CEO, Dario Amodei, has previously advocated for regulation to prevent harm from AI deployment and voiced considerable wariness about AI’s use in lethal operations and surveillance. This cautious approach has reportedly frustrated the US Department of Defense. The Secretary of War, Pete Hegseth, remarked in January that the department would not ’employ AI models that won’t allow you to fight wars,’ indicating a clear preference for less restrictive AI partners. The Pentagon is already collaborating with other AI companies, including xAI, Google’s Gemini, and OpenAI, for various research and operational support.
This unfolding story prompts crucial consideration: As artificial intelligence becomes more sophisticated, how should the international community balance military innovation with ethical considerations and the prevention of harm?