The Pentagon's cyber operations arm is forming a task force to accelerate the adoption of cutting-edge artificial intelligence tools with powerful hacking capabilities, according to three people familiar with the matter. The move from U.S. Cyber Command, which has not yet been publicly reported, highlights the Pentagon's concerns about the sudden emergence of private sector-developed artificial intelligence models that can find vulnerabilities in the security of digital systems faster than the world's top hackers.
Gen. Joshua Rudd, the dual leader of the National Security Agency and Cyber Command, announced the formation of the task force to employees two weeks ago in an internal email, two people familiar with the matter told POLITICO. According to the email, the working group will span Cyber Command and the National Security Agency and examine how the Pentagon can safely deploy leading artificial intelligence models across all aspects of its missions. This includes assessing how AI models developed by Silicon Valley tech giants can be used on "high-end" systems that carry the intelligence community's most sensitive secrets.
The formation of the task force comes as artificial intelligence giant Anthropic announced a month ago that its latest model, Claude Mythos, is so good at finding and exploiting network vulnerabilities that it will only be shared with a small group of trusted network defenders. Anthropic warned at the time that if the tool fell into the wrong hands, it could have "serious" consequences for the economy, public safety and national security. Other leading artificial intelligence companies, including OpenAI, have since announced they have models with similar advanced hacking skills and launched similar initiatives to restrict access. Anthropic officials predict that equivalent models will be widely available in six to 24 months, potentially enabling low-skilled hackers to wreak digital mayhem on a massive scale.
Concerns about the proliferation of advanced AI models have prompted emergency action from the White House, which is currently drafting an executive order requiring cutting-edge AI labs such as Anthropic, OpenAI and Google to submit their models to the federal government for testing before releasing them publicly. It marks a significant shift from the Trump administration, which has previously touted a hands-off approach to AI regulation.
While the task force is not directly tied to the upcoming executive order, Cyber Command and the National Security Agency are expected to play a central role in any new cybersecurity-related initiatives. The NSA, through its recently established Artificial Intelligence Security Center, is already working with the Department of Commerce to study the cyber capabilities of new cutting-edge AI models, an effort that may be expanded and codified in the new executive order. "The broader government is leaning on the NSA's technical expertise to figure out 'how do we do this safely,'" one person familiar with the matter said.
Any potential federal review would ensure that the U.S. intelligence community understands the capabilities of these new models before foreign adversaries such as Russia and China gain access. The NSA is responsible for intercepting foreign communications and protecting U.S. government security, while Cyber Command is responsible for defending against digital attacks on Department of Defense systems and supporting military operations.
In the email, Rudd said the task force would draw expertise from the National Security Agency's Artificial Intelligence Security Center and designated a Cyber Command commander to lead the task force. It's unclear how big the task force will be or how long it might operate. A former senior national security official said the creation of the task force reflects Rudd's belief that both Cyber Command and the National Security Agency need to deploy rapidly evolving artificial intelligence tools in their networks more quickly. The official added that assigning leadership roles to the commander of Cyber Command indicates "more dominance by the combatant commands, with NSA providing support."
However, much of the task force's technical prowess will likely come from the National Security Agency. The vast signals intelligence agency houses some of the top scientific and computer talent in the U.S. government, and Cyber Command has faced talent management problems for years. Earlier this month, the Pentagon announced that it had signed agreements with seven technology companies, including OpenAI and Google, to begin using its artificial intelligence models on classified networks.
Anthropic is currently locked in a legal dispute with the Pentagon, complicating wider deployment of Mythos across the federal government. The Defense Department took the unprecedented step in March of designating Anthropic as a supply chain risk after the company sought to limit the use of its tools in autonomous warfare and mass surveillance operations. An Anthropic official previously said the company was willing to allow the U.S. government to use Mythos in offensive cyber operations. This is a direction the Pentagon may explore, which could put pressure within the Department of Defense to lift Anthropic's supply chain risk designation.
Lt. Gen. Charles Moore, former deputy commander of Cyber Command, told POLITICO that he was "aware" of Cyber Command's plans to establish an artificial intelligence task force and said he viewed it as "not just a good idea, but a necessity." "Artificial intelligence tools are quickly becoming a necessity to detect threats, prioritize vulnerabilities, accelerate decision-making, and conduct defensive and offensive cyber operations faster than our adversaries," Moore said. "This working group can help integrate artificial intelligence into operations, training, intelligence and cyber defense while ensuring the United States maintains an analytical advantage over our adversaries."
Working with the private sector to improve the nation’s offensive cyber capabilities is at the center of the Trump administration’s new cybersecurity strategy released in March. In the strategy, the White House calls on federal agencies to adopt "artificial intelligence-driven cybersecurity solutions" to protect government networks. The strategy also says the Trump administration will "unleash the private sector" to help "expand our national capabilities," including cyber offense and defense.