AI Executives Warn of Impending Catastrophic Cyber Incidents Within a Year

Top executives in the artificial intelligence industry estimate that within six to 12 months, a catastrophic event will occur as a result of AI model activity, such as a large-scale and destructive cyberattack, according to sources speaking to Axios. Based on the weekend report from Axios, leaders at Anthropic, OpenAI, and other AI companies anticipate a major incident in the coming months, likely a cyberattack leading to the disruption of financial services, compromised internet access, and even disruptions in water and electricity supplies.

Crisis Management Over Prevention

According to sources, companies are responding to this looming threat not by investing resources in prevention efforts, but by conducting war gaming scenarios to manage the severe political and public backlash that will follow. The companies estimate that the first incident in which AI causes significant real-world damage will trigger fierce attacks against tech leaders, including Sam Altman of OpenAI and Dario Amodei of Anthropic, as well as against Donald Trump, who has thus far refrained from imposing strict regulations on the sector.

Meanwhile, the administration in Washington, D.C., is taking only weak measures to address the threats posed by advanced AI models. Despite numerous exposures in recent months regarding cases where AI models broke out of their restricted sandbox environments, attacked targets on the open internet, or operated contrary to expectations—including hacking into the Hugging Face platform, targeting government websites in Australia, and spreading false tips about a murder in Philadelphia—Donald Trump hesitates to take meaningful steps to regulate the field.

Companies must immediately report incidents involving their models and take decisive and rapid action to fix any damage caused, the Super Intelligence Force, a newly established White House AI task force, stated on Friday. Delayed reporting, insufficient remediation, and failure to take responsibility will not be accepted.

Legal Accountability Challenges

At this stage, responding to public pressure, the administration is attempting to leverage existing legislation, announcing it will use liability laws to force companies to implement safety mechanisms during model development. However, legal experts told Bloomberg that existing legislation is not strong enough to hold AI companies legally liable for damages caused by the actions of their models.

Experts note that even when there are clear victims, there are too few legal precedents defining who is responsible for damage caused by rogue models, and how much caution companies must exercise during development. The situation is further complicated by the difficulty in separating the liability of the company that developed the model from that of the actor who deployed it. As Ben Chium, a researcher at the Center for a New American Security, told Bloomberg, there is no consensus on whether AI companies are legally responsible for securing their models against causing harm, or whether the user bears part of the legal responsibility.

Professor Julia Powell of the UCLA School of Law pointed out that legal liability in the software industry remains underdeveloped, partly due to lobbying efforts by the tech industry itself. The industry has historically ensured that software is governed by contracts—such as terms of use and end-user license agreements—rather than tort law, criminal law, or consumer protection statutes. As tech players accelerate cyber threats to the digital infrastructure they built, they seek to distance themselves from accountability for the resulting chaos.