OpenAI Partners with US National Labs for Nuclear Safety and Research
OpenAI is collaborating with U.S. National Laboratories to advance research and enhance the security of nuclear weapons.
OpenAI Partners with the U.S. Government
In a significant development, OpenAI has declared its plan to provide its cutting-edge artificial intelligence models to around 15,000 scientists at U.S. national laboratories, signaling a new phase in its growing engagement with the federal government.
Collaboration with Microsoft
In partnership with Microsoft, OpenAI will deploy its o1 model, or another variant from the “o” series, on Venado. This supercomputer, recently unveiled by the Los Alamos National Laboratory, is powered by NVIDIA Grace Hopper.
Diverse Applications for AI Models
According to OpenAI, researchers at the Los Alamos, Lawrence Livermore, and Sandia national laboratories will use this technology to support various initiatives. These include protecting the national power grid from cyberattacks, discovering new disease treatments and prevention methods, and exploring fundamental physical laws. “OpenAI’s collaboration with U.S. national labs continues the longstanding tradition of partnership between the U.S. government and private industry to ensure that technological innovation leads to significant advancements in health, energy, and other critical areas,” the company stated.
Controversial Support for Nuclear Weapons
OpenAI also noted that its AI models would enhance work related to nuclear weapons, specifically supporting a program focused on reducing nuclear war risks and securing global nuclear materials and weapons. The company emphasized this support as “crucial” for its “commitment to national security”. OpenAI highlighted that its security-cleared researchers would provide a “careful and selective review of use cases and consultations on AI security”.
Prior to this announcement, the Los Alamos National Laboratory had already been using ChatGPT for various projects, such as exploring safe applications of AI models like GPT-4o in advancing bioscientific research.