The U.S. Department of Defense has awarded OpenAI a $200 million contract to bring generative AI to military systems, including command and control and elements of the nuclear triad. The company will “prototype cutting-edge AI capabilities to address critical national security challenges in both the warfighting and enterprise domains.”
The program with the U.S. Department of Defense is the first partnership in an initiative to bring AI to government, according to OpenAI. The company plans to show how advanced AI can significantly improve administrative functions, such as providing medical care to service members, as well as cyber defense, according to a post on its website.
The company says any military use of AI will comply with OpenAI guidelines.
Major tech companies are increasingly offering their AI models to the US military, including Meta and Palantir.
The new deal follows reports that senior executives at Palantir (Shyam Sankar), Meta (Andrew Bosworth), and OpenAI (Kevin Weil, Bob McGrew) have been promoted to lieutenant colonel in the U.S. Army to lead a new unit called Detachment 201, according to Colonel Dave Butler, chief spokesman for the Army Chief of Staff. Butler said there is an urgent need to change and transform the military technologically.
Team 201 members will work on specific projects for the U.S. Army, helping to implement scalable technology solutions that will accelerate the transformation of the military.
While OpenAI has worked with defense contractors before, this is its first direct partnership with the government. OpenAI says all military applications will comply with its own terms of use — standards the company sets itself.
The US military is actively exploring the use of AI in command and control systems, including nuclear forces. Although the official statement is that “the final decision will rest with humans,” the very formulation of the question raises concerns:
-
AI trained on historical data, may fail in non-standard situations, which in the nuclear sphere is fraught with fatal consequences.
-
Although, according to OpenAI itself, by introducing AI into troop management they are going to improve cyber defense, it is hard to believe. Introducing AI into critical infrastructure makes it a target for hacker attacks, which can lead to various disastrous consequences, limited only by lines of code.
-
Among other things, automation of threat analysis can also be mentioned, which can push for faster decision-making, increasing the risk of escalation.