Genesis is sold as AI for science. Its first challenges point straight at nuclear weapons

Genesis Mission’s – When the White House unveiled the Genesis Mission in November 2025, it framed the effort as an AI platform to accelerate research across science and technology. But the Department of Energy’s list of 26 challenges quickly shows a sharper focus: seven are expli

In November 2025, the White House rolled out a new AI program called the Genesis Mission. The promise was sweeping—an AI platform built with universities and private companies to accelerate research on problems too hard for today’s methods. But the Department of Energy. which oversees the nation’s nuclear stockpile. has now published 26 challenges for the effort. and seven of them land squarely on nuclear weapons and national security.

The contrast is uncomfortable for anyone who has learned to treat “AI for science” as a tidy label. On one side is the language of discovery: advanced manufacturing, biotechnology, nuclear energy, quantum information science, critical minerals, and semiconductors. On the other is a mission scope that includes protecting against nuclear threats. improving nuclear forensics. and helping the nuclear enterprise modernize faster.

By executive order last November. President Donald Trump tasked the Department of Energy with leading a “dedicated. coordinated national effort to unleash a new age of AI-accelerated innovation and discovery that can solve the most challenging problems of this century.” The mission is. in practice. intended to build an AI platform—partnering with universities and private companies—to tackle research in those broad areas. The DOE’s own public descriptions of Genesis have leaned heavily toward supernova research and disease modeling. emphasizing the “science” part of the story.

Then the challenges appeared.

Seven of the 26 challenges focus directly on nuclear weapons and national security. That makes sense. according to Bahrad Sokhansanj. a research scholar at the Institute for Law and AI in Boston who previously worked at Lawrence Livermore National Laboratory. Because of its nuclear-weapon research and its broader basic-science portfolio. he says the agency already has the infrastructure to run an AI-for-research effort at scale.

“It has labs, it has computers,” Sokhansanj says. “It has a lot of resources that are relevant to the future of science and technology.” (Lawrence Livermore, Los Alamos National Laboratory and DOE headquarters did not respond to requests for comment.)

Genesis, he adds, is a way to concentrate those resources on a single strategic objective rather than leaving them scattered across the government’s usual silos.

Still. folding AI deeper into nuclear work can feel unsettling—especially once the public hears “AI plus nukes.” Herbert Lin. a senior research scholar at Stanford University’s Center for International Security and Cooperation. argues the premise people fear is not what’s being proposed. He says the idea of AI taking charge of nuclear launches sits on one end of what he calls a spectrum.

People imagine a cinematic scenario: “an autonomous system deciding to launch an atomic attack.” Lin says nobody is proposing that arrangement. He points out that no one in the military, the National Security Council, or the acquisitions world wants it.

“Even the president, who is known for saying very nutty things, has not said that,” Lin says.

Lin frames the far more plausible range as everything between something benign—like using an AI grammar tool on a nuclear-policy memo—and the launch-code end of the spectrum: “There’s a whole range of stuff in between. ” he says. In that light, he says Genesis’s nuclear-AI plans “live somewhere between bad grammar and the launch codes.”.

The nuclear-oriented work Genesis describes doesn’t start with launching decisions. Several challenges are designed to protect people from nuclear threats by improving how systems detect and respond to attacks. One challenge would use AI to combine data from sensors. simulations. and intelligence to detect and respond to potential nuclear or radiological attacks.

Another aims to accelerate nuclear forensics—helping analysts characterize radioactive material recovered after an incident or from a weapon. trace where it came from. and determine whom to hold accountable. Genesis also includes a nonproliferation effort: using AI to analyze “satellite imagery. sensing. open-source and government data” to find anomalies that could suggest dangerous radioactive material moving somewhere it shouldn’t.

The mission also looks inward, toward maintaining and modernizing the nation’s arsenal. The DOE’s stockpile-stewardship program seeks to understand how aging nuclear weapons behave and how modernization efforts affect safety and performance. To do that. Genesis says it will use AI to mine the DOE’s archive of old data: results from classified experiments and weapons tests. along with unclassified nuclear research that may still be “trapped in hand-scrawled notes. paper files or film photographs.” Lin points to the possibility that the archive holds discoveries that have not yet been found.

“No doubt there’s interesting stuff in there they haven’t found,” Lin says. The system would also ingest and process new data.

For now, the new data in question will not come from nuclear weapon tests. But the DOE does study weapon behavior at facilities such as Los Alamos’s Dual-Axis Radiographic Hydrodynamic Test facility. which uses giant x-ray machines to make high-speed movies of imploding mock materials. One Genesis challenge would develop an automated system that behaves like an administrative layer—helping schedule experiments. steer them in real time. and run live diagnostics.

At high-hazard facilities where weapons modernization and production take place. Genesis would also use AI in safety planning for what outline documents describe as “streamlining production” and “removing red tape”—explicitly described as not the red tape that exists for good reason. Closer to the more sensitive end of Lin’s spectrum. one challenge is meant to tighten the handoff between the groups that design nuclear weapons and the groups that produce them. so that modernized systems can move ahead faster.

Those seven challenges are the most direct nuclear-security efforts. But Sokhansanj says the rest of Genesis—outside the headline nuclear list—still has national-security weight. The U.S. wants to “win” at AI while also pursuing self-sufficiency in areas such as critical minerals and manufacturing to reduce dependence on other countries. He links that idea to a longstanding reality: scientific leadership has operated as American soft power since the end of World War II. In the DOE’s world, he says, soft power tends to come with hard-power implications.

A key part of the story is dual use. Sokhansanj says. “A lot of the science the DOE wants to accelerate is dual use.” He points to examples: “If you’re working on particle accelerators. if you’re working on advanced factories. that’s all going to have other applications.” A particle accelerator. he says. can also help reveal how particles inside a weapon behave.

That raises governance issues. Sokhansanj adds. because the faster science moves. the more likely it is to produce results that could be dangerous as well as beneficial. He specifically points to synthetic biology research that could reveal “roadmaps to bioweapons.” In his view. Genesis should use some of its resources to fund countermeasures and cybersecurity. so that science secrets and breakthroughs aren’t stolen.

Lin’s warning lands on a different fear: the seductive idea that AI could serve as an all-knowing judge. He says some people think AI is an oracle that can tell you truth.

“[AI] can’t,” he says, explaining that this matters especially when evaluating actual or potential attacks.

There’s a clear through-line in Genesis’s first challenges. The program is being sold as AI for discovery—across supernovae. diseases. quantum information science. and more—but its published priorities show how quickly “discovery” becomes tightly entangled with nuclear risks. maintenance choices. and the governance choices that follow.

Genesis Mission AI for science Department of Energy nuclear weapons national security nuclear forensics nonproliferation dual-use research cybersecurity Los Alamos Dual-Axis Radiographic Hydrodynamic Test facility