Late last week, Avalanche reported that chief executives at Anthropic, OpenAI, and other AI firms are quietly considering scenarios that could spark a public uprising and political backlash in the wake of a severe AI-related incident. They foresee a major disruption—most likely a cyberattack—that could cripple financial services, disrupt the internet, and cut off electricity and water supplies. Insiders within the AI sector told the outlet that a significant event could occur within the next six to twelve months.
The industry fears that, should such a breach occur, a wary public would grow even more antagonistic toward the sector, blaming its leaders and the President for a perceived lack of government oversight. A recent poll conducted by the Associated Press and the NORC Center for Public Affairs Research found that 64 percent of American adults believe AI is advancing “too quickly.” The same survey indicated that 67 percent disapprove of Trump’s management of AI matters.
Despite the incidents that have already taken place, AI companies have faced accountability for their actions.
Following Australian Prime Minister Anthony Albanese’s disclosure last month that an OpenAI agent infiltrated a government website and accessed both public and nonpublic files, the company issued an apology and, as it has in other incidents, published a public repository detailing its findings. Yet Australian authorities continue pursuing OpenAI for potential violations of local laws. Attorneys who specialize in cyber and hacking law told TechCrunch that the fallout could range from federal hacking charges to civil litigation.
In the United States, courts may soon decide whether civil liability laws apply to the actions of rogue AI agents. In September, the nonprofit Legal Advocates for Safe Science and Technology filed suit against OpenAI under California’s Comprehensive Computer Data Access and Fraud Act, which prohibits knowingly accessing computer systems without authorization. While there is no clear evidence of malicious intent by the AI labs in the recent wave of rogue-agent cases, negligence is not excusable.
Axios points to a recent incident in which a Chinese actor used two AI tools—the open-weight model DeepSeek and Anthropic’s Claude Code—as an illustration of how blame could be directed toward AI labs. CrowdStrike, the cybersecurity firm that uncovered the hack, said the attacker targeted several South Korean banks and asked Claude for guidance on selling “Korean data breach information” and “finding Korean Telegram data sales groups.”
Bank hacks are not new, and the industry has generally acknowledged AI risks, yet public anxiety about these events has spurred U.S. lawmakers to consider far-reaching restrictions on the sector. Proposals include the AI Kill Switch Act, a bipartisan bill that would empower 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, operate, and train AI models; and a recently proposed plan to fund new social welfare programs through taxes on data centers and AI firms.
It appears that Trump, who has dismissed calls for regulation of the industry, is still shaping a path for federal involvement in AI. Over the weekend, he announced the creation of the Super Intelligence Force, described as a body “tasked with coordinating the effort of the Federal Government to ensure that America continues to lead the World in Super Intelligence.” The president asserted that the force would safeguard the interests and improve the lives of all Americans and would coordinate with a broad array of organizations. Notably, Congress—the branch most responsive to public pressure and the sole body empowered to draft and pass laws—was not listed among the participants.
According to the group’s charter, which Politico obtained, the scope of the Super Intelligence Force appears to be limited to producing reports and making recommendations rather than enacting policies.
These developments suggest the federal government may eventually adopt a heavy-handed approach to AI regulation. “One Democratic aide,” speaking with Axios, pointed to the cooperation seen during COVID-19 and the 2008 financial crisis as evidence that government can set aside partisanship in the face of catastrophe. Yet those crises also demonstrate why many resist giving Washington more sway over an industry it does not fully understand.
While the idea that AI firms are preparing for worst-case scenarios may be unsettling, it constitutes prudent risk management and aligns with the industry’s pledge to engage in responsible self-policing. After all, these companies have a powerful incentive to plan for extreme outcomes: if a doomsday event unfolds, the customer base could vanish.
