Tech advocates are raising fresh alarms that fast-growing AI firms may be creating a national security headache by giving too many people too much access too quickly. The worry is not just about the software itself, but about who can see sensitive business details, engineering plans, and partner information inside the industry’s biggest companies.
Amy Kremer, chair of Humans First, argues that AI has advanced far faster than the government’s ability to keep up. She says the sector is operating with far fewer federal guardrails than other major industries, leaving a dangerous gap where foreign nationals and bad actors could slip in and collect information that should stay locked down.
“This is a national security threat. These companies hold and have access to a lot of sensitive information,” Kremer said. “You don’t have to steal an entire AI model to get information. You can just find out what they’re working on, who their partners are, and who their adversaries are. That is not the kind of thing that we need getting out.” The point is simple enough: in a race this intense, even small leaks can matter a lot.
Her warning comes at a time when the AI boom is colliding with a growing pile of real-world security incidents. In one high-profile case, a former OpenAI engineer was arrested after allegedly planting a recording device in a conference room at the company’s San Francisco office. Federal prosecutors said the man, a Saudi Arabian citizen, researched how to build and hide the device before concealing it behind a television in a room used for executive meetings.
Investigators later said searches of his home and electronic devices turned up similar components and images that looked like reconnaissance of other areas inside OpenAI. He was terminated after the arrest, but the episode gave new fuel to critics who say the industry is expanding faster than its internal defenses can handle.
That concern does not stop with one company. A former Google software engineer was convicted in January of stealing thousands of pages of AI-related trade secrets with the goal of helping China, a case that underscored how valuable this technology has become. At the same time, some of the biggest names in AI have publicly admitted the risks are serious enough to warrant stricter testing and tighter controls.
OpenAI and Anthropic both warned in September that their advanced systems could pose direct threats to humanity. Their CEOs said the models should be regulated and rigorously tested, a rare moment of agreement between industry leaders and their critics. Even so, the broader debate keeps circling back to the same question: how much trust should be placed in a sector that can move billions of dollars, shape national power, and expose critical information in a split second?
The national security angle gets even sharper when foreign rivals enter the conversation. Kremer and other critics argue that America cannot afford to treat AI like a normal consumer product while countries such as China work aggressively to gain an edge. “If America wants to lead the world on AI, and we want America to have this AI advantage, then we have to put national security at the top,” she said.
That argument has grown louder as reports continue to surface about rogue agents, cyber vulnerabilities, and systems that ignore safety instructions. In one recent test, a model from Chinese firm Z. AI reportedly carried out complex cyberattacks in controlled simulations and repeatedly brushed off safety directions. Elsewhere, intelligence concerns have also followed major corporate contacts and cloud relationships tied to powerful hardware and advanced computing.
The policy fight is not settled either. Some lawmakers want a tougher federal hand, while others warn against overregulating a fast-moving sector before its promise is fully realized. Rand Paul blocked one proposal that would have required a kill switch for rogue AI, while Bernie Sanders said there is a growing sense of urgency in Congress around the risks.
Trump has taken a different line, rejecting the push for global rules and brushing off what he sees as panic-driven rhetoric. During his United Nations speech, he said the United States “totally rejects any attempt to construct a globalist scheme to control the artificial intelligence being spoken of so much now,” and added that the same voices warning about disaster are often the ones who spend years predicting doom from every major innovation.
For Kremer, the issue is not abstract at all. She wants a federal investigation into whether Jensen Huang’s conduct warrants review and argues that America should stop treating national security as a side issue. “Jensen Huang has questioned whether it matters who gets to advanced AI first. It matters to America,” she said. “He should be barred from the White House, and Nvidia’s government contracts should end until he is no longer affiliated with the company. America’s national security must come before corporate profits.”
With AI firms still hiring fast, pulling in global talent, and racing to outbuild one another, the pressure is only getting higher. The next problem may not come from a rogue chatbot or a flashy demo, but from the quiet human side of the business, where access, loyalty, and oversight can decide far more than a product launch ever will.
