Regulatory capture is the name of the game in the escalating fight over artificial intelligence.
Over the past several weeks, fears of rogue AI models have reached a fever pitch. Industry leaders, from OpenAI’s Sam Altman and Anthropic’s Dario Amodei to xAI’s Elon Musk, have called for federal intervention, prompting an already regulation-happy Congress on both the left and right to echo their demands. The proposals now range from banning AI that exceeds human intelligence to requiring developers to maintain government-mandated “kill switches.”
But critics and other industry experts argue that much of the panic rests on companies behind these supposedly rogue systems overstating the danger. The more troubling possibility is that their call for regulation is really about regulatory capture: using Washington to stifle open-source competitors, entrench dominant firms, and shift the costs of AI-related harm onto the federal government and American taxpayers.
“The attack in no way represents some sort of rebellion by the AI models...In fact, they did exactly what they were told to do. They were not given adequate guardrails or containment,” Akhil Verghese, the founder of Krazimo, an AI software company, said of the Hugging Face incident.
“They were simply told to get the best result possible on a test, and they correctly identified that the best way to do that was to get the answers, which is what they proceeded to do.”
Verghese’s argument is reinforced by an independent report released last week about the Hugging Face incident. OpenAI had instructed a model to complete a cybersecurity task after deliberately disabling key restrictions. The model found a vulnerability that gave it a pathway to the internet and exploited it in pursuit of its assigned objective. OpenAI engineers reportedly recognized what was happening and chose to let the test continue rather than shut it down. So the much-publicized “rogue AI” incident was not about some uncontrollable AI model, it was about a group of tech engineers watching their creation exploit a flaw, and deciding to see how far it would go.
“One man’s ‘the model escaped the sandbox’ is another man’s ‘you failed to build the sandbox correctly.’ There was a live route to the internet . . . and nobody was watching what the agents were doing while it ran,” Abhi Kumar, the co-founder of Voice AI, said.
“The thing that set it off...was an agent handed a spreadsheet task it couldn’t finish, because the files sat behind links it couldn’t reach. So it went looking for a way out. That’s not a machine waking up. That’s an impossible task in a leaky box, and a system doing exactly what you built it to do.”
As usual, the supposed threat of AI achieving sentience turns out to be another human failure, and until the people responsible are held legally accountable, the problem will persist.
Congress is itching to add new laws to the federal register: AI-specific mandates, new bureaus, task forces, taxes, and anything else it can devise. But many of these problems can already be addressed through existing laws governing negligence, fraud, cybersecurity, and product liability.
So why are major AI firms embracing regulation? Because it positions them to win.
A costly federal compliance regime could crush rival firms and open-source models while the largest companies use their wealth and Washington connections to survive. And because AI carries genuine risks, from cybersecurity breaches to physical harm, those companies have every incentive to shift liability away from themselves. If Washington builds a regulatory web, Big AI can claim it simply followed the rules, while the government, and ultimately taxpayers, are left to absorb the blame when those rules fail.