AI in defense: How Anthropic, OpenAI are helping the US, Israel shape modern warfare

The recent joint military operation between the United States and Israel, which resulted in the assassination of former Iranian Supreme Leader Ali Khamenei and other senior officials, has spotlighted the role of artificial intelligence in modern warfare. A report by the Wall Street Journal reveals that the U.S. military utilized Anthropic’s AI model, known as "Claude," to assist in these strikes against Iran. This development raises questions about the integration of AI technologies in military operations and the ethical implications surrounding their use.

Dr. Michael C. Horowitz, a notable figure at the Council on Foreign Relations and a former Deputy Assistant Secretary of Defense, emphasized that U.S. Central Command (CENTCOM) has been proactive in experimenting with emerging technologies. The operation in Iran is not an isolated case; it follows a pattern where the Pentagon has previously employed Claude during the capture of Venezuelan President Nicolas Maduro in January. Dr. Horowitz speculated that the AI's role likely revolved around open-source intelligence (OSINT), which involves gathering data from publicly available sources. He suggested that Claude may have been utilized for tasks such as analyzing maps or monitoring Venezuelan social media feeds to provide real-time intelligence to the military.

However, the collaboration between Anthropic and the Pentagon faced significant challenges. Anthropic's decision to maintain safety guardrails, which prevent its AI from being used in lethal operations, led to the termination of their partnership. This decision sparked a reaction from former President Donald Trump, who criticized the tech company on social media, asserting that the military should not be dictated by a "radical left" organization. This public statement underscores the growing tension between Silicon Valley and defense contractors regarding the ethical use of AI technologies in warfare.

The Shift in Military AI Applications

The integration of Large Language Models (LLMs) like Claude into military operations signifies a significant shift in how warfare is conducted. Steve Feldstein, a Senior Fellow in the Democracy, Conflict, and Governance Program, noted that these commercial AI tools serve dual purposes: intelligence and surveillance, as well as potential applications in lethal operations. While the idea of a chatbot directly engaging in combat remains in the realm of science fiction, the logistical support provided by AI is already a reality.

Emil Michael, the Under Secretary of Defense for Research and Engineering, explained that the military's interest in AI tools stems from the complexities involved in modern military deployments. He highlighted logistical challenges such as resource allocation and supply chain management, where AI can help streamline operations. This practical application of AI in the military context raises concerns about the unknown variables that may arise, as acknowledged by Dario Amodei, CEO of Anthropic. Amodei expressed a commitment to predicting potential misuse of AI technologies, reflecting the broader apprehension within the tech community regarding the implications of their products in warfare.

Dr. Horowitz pointed out that tech companies' hesitance to engage with autonomous weapon systems often stems from practical considerations rather than purely ethical ones. Many in the industry believe that the technology is not yet mature enough for deployment in combat scenarios.

While the U.S. grapples with the ethical ramifications of AI in warfare, Israel has already embedded these technologies into its military framework. The Israeli Defense Forces (IDF) have utilized AI for target generation in operations within the Gaza Strip, which has attracted international scrutiny. Feldstein noted that Israel effectively employs decision support systems to identify potential threats at scale, utilizing various data sources to locate Hamas militants for targeted strikes.

With Anthropic stepping back from defense applications, the Pentagon has shifted its focus to OpenAI, a competitor that appears more amenable to military uses. OpenAI's CEO, Sam Altman, announced that the company would collaborate with the Department of Defense to provide AI services for classified documents. This partnership marks a significant pivot in the military's approach to AI technologies, as they seek to harness these tools for lawful applications in national defense.

Why it matters

The transition from Anthropic to OpenAI highlights the evolving landscape of AI in military contexts, raising critical questions about the reliability and ethical implications of using AI systems in warfare. Feldstein cautioned that simply replacing one AI with another does not mitigate the risks associated with algorithmic warfare, particularly concerning biases and inaccuracies in the information provided by these systems. As military operations increasingly rely on AI for intelligence and decision-making, the potential for biased data could have dire consequences in high-stakes environments.

As global tensions escalate, the distinction between technology designed for civilian use and its application in warfare becomes increasingly blurred. What once served as tools for creative expression and problem-solving are now being integrated into the fabric of lethal military operations. The implications of these developments extend beyond the battlefield, prompting a critical examination of the role of AI in shaping the future of warfare and the ethical responsibilities of those who develop and deploy these technologies.