OpenAI, initially reluctant to any military application of its models, has announced a partnership with Anduril for defense systems. Does this collaboration pave the way for combat drones and AI-optimized weapons?
Anduril, an American weapons company, has partnered with OpenAI, the creator of ChatGPT. The stated goal is to improve American defense systems using AI. OpenAI, which previously refused any military use of its models, is therefore changing its strategy.
“Our partnership with Anduril will help protect American military personnel and will help the national security community use this technology responsibly to ensure the safety and freedom of our citizens”, said Sam Altman, CEO of OpenAI.
A partnership truly limited to defense systems?
The announcement seems, at first glance, reasonable. The partnership mainly aims to use AI to analyze real-time data and detect aerial threats, whether planes or drones. A "passive" use, supporting human teams without actively participating in combat.
The press release also highlights the AI competition between the United States and China. However, this partnership raises concerns: Anduril is not limited to detection systems. Specializing in interconnected autonomous missile swarms and combat drone swarms, the company could go much further.
Such partnerships are multiplying between AI giants and defense companies. Anthropic is collaborating with Palantir, and Meta has announced a partnership for its Llama model with several companies, including Anduril and Palantir. If current announcements concern defense, this probably only represents a first step. These weapons companies now have access to the best language models and could develop fully automated offensive systems.