[US-Iran War] Anthropic Used in Iran Airstrikes...Public Focus Shifts to 'AI Ethics'

Trump Criticizes Anthropic as a "Leftist Company"

Anthropic's "Claude" Used to Identify Targets

Rising Interest in AI Ethics... "Human Rights Must Be Prioritized"

The Trump administration's use of Anthropic's AI model in successive military operations has heightened public attention on AI ethics.

On March 2 (local time), the international community expressed concern about the reality that AI technology is being used as a core strategy in warfare, following a report by the Wall Street Journal (WSJ) that the United States utilized Anthropic's AI model, Claude, in its airstrikes on Iran. While the U.S. Department of Defense and the Central Intelligence Agency (CIA) did not disclose specific details on the scope of use, it is known that the United States Central Command (CENTCOM), which oversees U.S. military operations in the Middle East, leveraged Claude for intelligence gathering, target identification, and combat scenario simulations. Additionally, Claude reportedly assisted in identifying high-value targets such as members of the Iranian leadership and in establishing attack priorities.

The airstrike on Iran utilizing Claude occurred just hours after U.S. President Donald Trump declared that the federal government would not use Anthropic’s products. President Trump and Anthropic have clashed over the use of AI in the defense sector since last year. Tensions reached a peak in January when it was revealed that the U.S. Department of Defense used Claude in the process of apprehending Venezuelan President Nicolas Maduro.

Anthropic has refused the U.S. Department of Defense's request to deploy its model on classified networks, stating, "Our models cannot be used for domestic surveillance or fully autonomous weapons systems that exclude human final decision-making." The Department of Defense has pressured Anthropic by invoking the Defense Production Act (DPA), which allows the government to control private industry production for national security. President Trump also criticized Anthropic via his social media platform, Truth Social, labeling the company as "a radical left-wing enterprise."

Anthropic Born from a Call for “AI Control”… Claude Tops App Store Rankings

Anthropic’s strong stance on defense-related issues is connected to its origins. Dario Amodei, Chief Executive Officer (CEO) of Anthropic, joined OpenAI in 2016 and played a key role in developing ChatGPT. However, in 2021, internal conflict erupted at OpenAI over the pace of commercialization and its pursuit of profit. CEO Amodei, who believed in the need for AI control, left OpenAI and founded the new AI company Anthropic.

Public opinion is increasingly valuing AI companies that prioritize ethical standards. Claude, which has taken an anti-Trump stance while emphasizing AI ethics, is a prime example of this soaring popularity. On February 28, Claude surpassed OpenAI’s ChatGPT to claim the number one spot among free apps on the U.S. Apple App Store, causing access errors due to a surge in usage. In contrast, OpenAI has faced heavy public criticism after signing a contract to provide its services to the U.S. Department of Defense’s classified networks. Hakim Jeffries, Democratic House Minority Leader, pointed out, "Mass surveillance of American citizens is unacceptable."

Interest in AI ethics is also growing in South Korea. On February 25, People's Solidarity for Participatory Democracy and the Institute for Peace and War Studies at Sungkonghoe University held a forum titled "Artificial Intelligence and the Future of War" at the Neutinamu Hall of People's Solidarity for Participatory Democracy in Jongno-gu, Seoul. Paek Beomseok, professor at Kyung Hee University Law School, stated, "Entrusting decisions that could cause large-scale civilian casualties to AI presents a serious moral problem," and emphasized, "Human rights must be prioritized in every phase from technology design to decommissioning."

This content was produced with the assistance of AI translation services.