America’s top tech leader in the War Department is letting Silicon Valley know it’s time to get serious about national security.
Emil Michael, the undersecretary for research and engineering, didn’t mince words when he said AI companies have a duty to stop pretending their creations are neutral tools.
Some of these systems, he warned, are potential “cyber weapons” — and companies can’t keep shrugging off the responsibility that comes with unleashing them.
Speaking at the Washington Post’s inaugural Building America Summit, Michael made clear that the Pentagon won’t sit back while private tech firms chase profits and ignore the national security consequences of their innovations.
He pointed directly at Anthropic’s new AI system, named Mythos, as one of those risky models sitting at the edge between innovation and catastrophe.
“These companies have a responsibility to ensure that their weapons, what they call weaponization potential of these models, to be careful and thoughtful about what they’re doing,” Michael said.
His point: AI isn’t just about fun apps or clever chatbots — it’s already part of the global battlefield, whether companies like it or not.
His comments come in the wake of President Donald Trump’s latest executive order on AI, which pushes innovation while reminding the tech giants that national defense takes precedence over their corporate virtue signaling.
The order creates a new “AI cybersecurity clearinghouse” where AI companies can voluntarily let the government scan their software for vulnerabilities before release. It’s voluntary, for now — but the message is loud and clear.
Michael noted that firms with “weaponization capability” should allow government experts to spend 30 days examining their systems.
Such evaluation could expose security gaps in critical U.S. infrastructure like power grids, water systems, and hospitals — the exact places America’s enemies would target if those AI tools fell into the wrong hands.
“I think they’ve all agreed and think it’s a good idea to do that,” Michael explained.
“I give all the companies, OpenAI, even Anthropic, and Google credit for sort of agreeing that was a smart thing to do.” It’s a rare thing these days to see massive tech players actually align with the Pentagon’s sense of duty — but apparently even Silicon Valley knows it can’t afford to get on the wrong side of President Trump’s national security revival.
One company that remains a thorn in the Pentagon’s side is Anthropic. The firm has been left out of several major defense partnerships after refusing to allow unrestricted access to its Claude models for military applications.
Anthropic even sued the Trump administration after being labeled a supply chain risk — a move that only deepened the perception that the company prioritizes ideology over American security.
Now, Anthropic’s new product, Mythos, is drawing fire from cybersecurity experts who see it as a hacker’s dream tool.
The company itself bragged that the system can “find ways to exploit vulnerabilities” in software, essentially advertising the fact that their model can be weaponized by any adversary with a keyboard.
Critics inside the intelligence and defense communities are warning that selling such technology without guardrails is an open invitation to foreign cyberattacks.
Meanwhile, the War Department isn’t sitting idle. AI has already been deeply integrated across the department’s operations, from intelligence analysis to logistics optimization.
“Six months ago, only about 80,000 federal employees were AI users each month. Now, there are 1.5 million,” Michael said. The government, he added, has “raced” to push AI deeper into daily workflows to gain major advantages in efficiency, intelligence, and warfighting readiness.
By year’s end, Michael expects that three-quarters of the entire department will be using AI systems in their duties. “We’ve integrated all the biggest AI companies over the last few months,” he said, adding that this surge in adoption means “we’re, in one year, going to make progress more than the five years before it.”
The push marks another step in President Trump’s ongoing effort to accelerate modernization inside America’s national security infrastructure.
Under War Secretary Pete Hegseth and leaders like Michael, the Department of War is rebuilding a culture that emphasizes strength through innovation — not bureaucratic red tape. The goal isn’t just technological advancement; it’s deterrence through superiority.
Michael’s call for corporate accountability in AI development reflects a broader understanding that warfare now extends far beyond physical battlefields.
Cyber capabilities, machine learning, and automation are now weapons of equal — if not greater — strategic value. If private companies aren’t responsible, the consequences could be disastrous, both for them and for the nation they rely on to keep them free to operate.
The Pentagon’s warning should not be ignored. The next generation of American defense will depend on how well AI is harnessed — and how securely it’s built.
The companies that take that duty seriously will find themselves on the winning team. Those that don’t may find themselves locked out of the future of national defense entirely.