OpenAI partners with US National Laboratories on research and nuclear weapon safety

Just days after announcing a version of ChatGPT designed for US government use, OpenAI is further entangling itself with the federal government. The company announced Thursday it would provide approximately 15,000 scientists associated with the US National Laboratories access to its latest frontier AI models. OpenAI will work with Microsoft to deploy its o1 model “or another o-series” variant on Venado, the Los Alamos National Laboratory’s recently unveiled NVIDIA Grace Hopper-powered supercomputer.

According to OpenAI, researchers from the Los Alamos, Lawrence Livermore and Sandia National Labs will use the company’s technology to assist work they’re doing to protect the national power grid from cyberattacks, identify new ways to treat and prevent diseases and study the fundamental laws of physics, among other initiatives.

“OpenAI’s collaboration with the US National Labs builds on the long-standing tradition of the US government collaborating with private industry to ensure that technological innovation leads to meaningful improvements in healthcare, energy, and other critical fields,” the company said.

Perhaps controversially, OpenAI says its AI models will also enhance work involving nuclear weapons — specifically a program “focused on reducing the risk of nuclear war and securing nuclear materials and weapons worldwide.” According to the company, this support is “critical” to its “commitment to national security.” It adds that OpenAI researchers with security clearance will offer “careful and selective review of use cases and consultations on AI safety.”

Before today, the Los Alamos National Laboratory was already using ChatGPT. For instance, one of the lab’s divisions is studying how AI models like GPT-4o could be safely used to advance bioscientific research. More broadly, federal, state and local government workers at 3,500 agencies across the country have sent more than 18 million messages to the chatbot since 2024, according to OpenAI.

This is the latest move by OpenAI apparently intended to curry favor with the Trump administration.. Last week, OpenAI announced it was partnering with SoftBank to build $500 billion worth of AI infrastructure across the US. Before that, Altman personally contributed $1 million to President Trump’s inauguration.