"Offering Up to $600,000 a Year": Big Tech Recruits Weapons Experts to Build AI Ethical Safeguards

Anthropic, OpenAI, Microsoft, Google, and Meta
AI technology mobilized for warfare
Recruiting weapons experts to establish basic safety safeguards

U.S. big tech companies are actively recruiting weapons and biochemical experts. Their aim is to establish ethical safeguards, such as designing guidelines to prevent artificial intelligence (AI) models from generating dangerous chemical information.

According to the information technology (IT) industry on March 25, Anthropic is currently hiring a policy manager specializing in chemical weapons and high-performance explosives. The company is seeking Ph.D.-level experts with more than five years of experience in weapons defense, offering an annual salary in the range of approximately $245,000 to $280,000 (about 360 million to 410 million won). Anthropic is also recruiting CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosives) technical threat investigators under similar compensation terms.

This recruitment initiative by Anthropic is intended to "prevent the lethal misuse of AI systems." Given the company's history of conflicts with the U.S. Department of Defense over the expansion of AI's military applications, Anthropic seeks to build an ethical defense line to ensure its AI model, Claude, is not used for lethal weapon development.

OpenAI, which has recently strengthened its collaboration with the Department of Defense, is also developing safety systems to prevent AI from causing risks related to nuclear and chemical weapons. While the company previously removed a clause prohibiting the use of AI for "military and warfare" purposes, it continues to limit permissible uses to areas such as enhancing cybersecurity and optimizing transport logistics. OpenAI maintains the principle of not allowing its technology to be used for harming people or developing weapons. In relation to this, the company is hiring biochemical threat analysts and other specialists within teams dedicated to identifying and managing AI model risks. Compensation for these roles is higher than industry averages, at around $300,000 to $450,000 per year (about 440 million to 670 million won).

For similar reasons, Microsoft (MS), Google, and Meta are actively seeking such experts as well. MS, the largest investor in OpenAI and a developer of its own AI models, has reportedly been operating a specialized red team since 2023. The team tests AI to ensure it cannot provide instructions on manufacturing biochemical weapons or synthesizing controlled substances.

Google is considered one of the companies most sensitive to biological risk management, as it operates AlphaFold, an AI system for protein structure prediction. Google currently has a dedicated biosecurity team and is working with Anthropic and others to establish safety standards for biochemical weapons. Meta, which releases its models as open source, has to be even more attentive to pre-training. The company is hiring a large number of biochemical experts who can pose thousands of risky questions to the model before its release and train it to refuse to answer.

Lee Sangkeun, a professor at Korea University’s Graduate School of Information Security, commented, "In times of war, the use of technology for survival inevitably takes precedence over ethics. It appears that companies are trying to maintain safety lines even as they are swept up by the situation." He added, "However, as offensive and defensive technologies tend to develop together, caution is necessary. It seems that Korea is taking a cautious approach."

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