The Pentagon’s cyber-warfighting arm is launching a task force to speed up the adoption of cutting-edge artificial intelligence tools with powerful hacking capabilities, according to three people with knowledge of the effort.
The initiative from U.S. Cyber Command — which has not been previously reported — underscores the Pentagon’s concerns about the sudden emergence of private sector-built AI models that can unearth security flaws in digital systems faster than the world’s best hackers.
The task force was announced to staff two weeks ago by Gen. Joshua Rudd, the dual-hat leader of the National Security Agency and Cyber Command, according to an internal email described to POLITICO by two of the people with knowledge of the effort.
These people, like others in this report, were granted anonymity because they were not authorized to speak publicly about the sensitive effort.
According to the email, the task force will span Cyber Command and the NSA, and will study how the Pentagon can safely deploy leading AI models in all aspects of its missions, the two people said. This includes assessing how AI models built by Silicon Valley tech giants can be used on “high-side” systems bearing some of the intelligence community’s most sensitive secrets.
Spokespeople for Cyber Command, the NSA and the Pentagon did not respond to a request for comment.
The creation of this task force comes a month after AI giant Anthropic announced that its newest model, Claude Mythos, was so adept at finding and exploiting cyber vulnerabilities that it would be shared only with a small group of trusted cyber defenders. Anthropic warned at the time that the fallout across economies, public safety and national security could be “severe” if the tool were to fall into the wrong hands.
Other leading AI firms, including OpenAI, have since announced they have models with similarly advanced hacking skills and launched their own such initiatives to limit access. Anthropic officials have projected that equivalent models will be widely available in six to 24 months, potentially allowing low-skilled hackers to sow widespread digital disruption.
Concern over the proliferation of advanced AI models has sparked a frantic effort by the White House to cobble together an executive order that would have frontier AI labs such as Anthropic, OpenAI and Google submit their models for federal testing before public release. It marks a major shift for the Trump administration, which previously touted a hands-off approach to AI regulation.
The task force is not directly connected to the forthcoming EO, although both Cyber Command and NSA are expected to play a central role in any new cybersecurity-related efforts.
Through its recently established AI Security Center, the NSA already works with the Commerce Department to study the cyber capabilities of new frontier AI models — work that could be expanded and codified in the new EO.
“The broader government is leaning on NSA tech expertise to say, ‘how do we do this safely?’” said the first of the two people.
POLITICO previously reported that any potential federal vetting regime would ensure the U.S. spy community can understand the capabilities of these new models before foreign adversaries like Russia and China get access to them.
The NSA is responsible for intercepting foreign communications and securing the U.S. government’s, while Cyber Command is charged with fending off digital attacks on Defense Department systems and supporting military operations.
In his email, Rudd indicated the task force would draw expertise from the NSA’s AI Security Center, and identified a commander from Cyber Command to lead it, said the two people. Both declined to name the commander for security reasons.
It is not clear how large the task force is or how long it could be in operation. The Cyber National Mission Force, Cyber Command’s operational wing, was organized into six joint task forces as of 2024, and historically has supported Cyber Command’s work on election security, ransomware attacks and other cyber threats.
A former senior national security official, granted anonymity for fear of retribution, said standing up the task force indicates Rudd felt both Cyber Command and NSA need to move more quickly to deploy fast-advancing AI tools across their networks. This person added that assigning the lead role to a commander at Cyber Command suggested that the combatant command “is more in charge, with NSA in support.”
Most of the technical horsepower for the task force, nonetheless, is likely to come from NSA, the person added. The sprawling signals intelligence agency boasts the top scientific and computer talent in the U.S. government, while Cyber Command has for years faced talent management issues.
Earlier this month, the Pentagon announced it had signed deals with seven tech companies, including OpenAI and Google, to start using their AI models on classified networks.
Anthropic is locked in a legal battle with the Pentagon, complicating the wider rollout of Mythos across the federal government. The Defense Department in March took the unprecedented step of designating Anthropic a supply chain risk, after the firm sought to restrict the use of its tools in autonomous warfare and mass surveillance operations.
An Anthropic official previously told POLITICO the company is open to allowing the U.S. government to use Mythos in offensive cyber operations. That is something the Pentagon is likely to explore — and it could create pressure within the Defense Department to lift the supply chain risk designation on Anthropic.
Lt. Gen. Charles Moore, former deputy commander of Cyber Command, told POLITICO he was “aware” of Cyber Command’s plans to stand up the AI task force, and said he viewed it “not only as a good idea but a necessity.”
“AI tools are rapidly becoming essential for detecting threats, prioritizing vulnerabilities, accelerating decision making and conducting both defensive and offensive cyber operations faster than our adversaries,” Moore said. “The Task Force can help integrate AI across operations, training, intelligence and cyber defense while ensuring the United States maintains analytic superiority over our adversaries.”
Working with the private sector to boost the nation’s offensive cyber capabilities is core to the Trump administration’s new cybersecurity strategy, published in March. In it, the White House called for federal agencies to adopt “AI-powered cybersecurity solutions” to protect government networks. It also said the Trump administration would “unleash the private sector” to help “scale our national capabilities” both on cyber offense and defense.
