Don’t Blame AI Labs for What Hackers Do With Their Tools

On Friday, Axios reported that “Top executives at Anthropic, OpenAI and other AI companies are privately gaming out scenarios for a public and political revolt after a catastrophic AI event.” The executives “anticipate a large-scale event, most likely a cyberattack,” that would shut down financial services, disrupt the internet, and cut off power and water. Insiders in the AI industry told the outlet that a “major event will occur in the next six to 12 months.”

AI companies are rightly worried that if a major hack were to happen, “an already wary public,” according to Axios, would turn even more against the industry and blame its leaders and President Donald Trump for a presumed absence of government oversight. A poll by the Associated Press and the NORC Center for Public Affairs Research released Thursday found that 64 percent of U.S. adults “think AI is developing ‘too fast.'” The public also doesn’t think highly of Trump’s handling of the industry, with 67 percent of U.S. adults “disapproving of his performance on AI.” 

With the incidents that have already occurred, AI companies have been held accountable.

After Australia’s Prime Minister Anthony Albanese revealed last month that an OpenAI agent infiltrated an Australian government website and accessed public and nonpublic files, the company issued a mea culpa and, as it has in multiple incidents, created a public repository of its findings. That hasn’t stopped Australian law enforcement from trying to hold OpenAI accountable for any violations of the country’s laws. Attorneys specializing in computer and hacking laws told TechCrunch that the “potential fallout ranges from federal hacking charges to civil litigation.” 

In the U.S., courts could soon decide if civil litigation laws apply to rogue agents’ actions. In September, the nonprofit Legal Advocates for Safe Science and Technology sued OpenAI under California’s Comprehensive Computer Data Access and Fraud Act, which prohibits knowingly accessing computer systems without authorization. No evidence shows malicious intent by the AI labs in the recent spate of rogue AI agent actions, but that doesn’t excuse negligence. 

In its reporting, Axios points to a recent hack by a Chinese actor using two AI tools—the Chinese open-weight model DeepSeek and Anthropic’s Claude Code—as an example of what could lead to blame falling on AI labs. CrowdStrike, the cybersecurity firm that uncovered the hack, said the hacker targeted several South Korean banks and asked Claude for advice on selling “Korean data breach information” and “finding Korean Telegram data sales groups.” 

While bank hacks aren’t new and the industry has been pretty transparent about AI’s risks, the public’s anxiety about these incidents has inspired U.S. lawmakers to propose looming restrictions on the sector. This includes ideas such as the AI Kill Switch Act, a bipartisan bill that would allow the federal government to shut down powerful AI systems; the Artificial Intelligence Data Center Moratorium Act, which would halt construction of new data centers that host, run, and train AI models; and a recently proposed plan to fund new social welfare programs with taxes on data centers and AI companies.

It appears Trump, who has shot down calls to regulate the industry, is still carving out a lane for the federal government to shape AI. On Sunday, Trump announced the formation of the Super Intelligence Force, a group “tasked with coordinating the effort of the Federal Government to ensure that America continues to lead the World in Super Intelligence.” The president said the force will “protect the interests, and improve the lives, of all Americans,” and coordinate with a slew of organizations and entities. The list did not include Congress, the branch most responsive to public pressures, and the only body with the power to draft and pass legislation.

According to the group’s charter, which Politico obtained, the Super Intelligence Force’s scope appears limited to reports and recommendations.

These actions suggest the federal government will eventually take a heavy-handed approach to regulating the industry. “One Democratic aide,” in conversation with Axios, cited the “cooperation during COVID and the 2008 financial crisis” as evidence that government could put partisanship aside in the face of catastrophic events. Yet the government’s response to those crises is even more reason to resist calls for Washington to swing its weight around in an industry it doesn’t really understand. 

Hearing that AI firms are preparing for worst-case scenarios might be chilling, but it’s good risk management and the kind of self-policing the industry has pledged to implement. After all, these companies have a glaring incentive to plan for worst-case scenarios: There’ll be no customers after doomsday.