Zywave Cyber Front Page News
- Saturday, October 10, 2026
AI 'near-misses' show potential losses beyond cyber: CyberCube
AI 'near-misses' show potential losses beyond cyber: CyberCube
By Erin Ayers, Front Page News
As the insurance industry tries to more fully understand the liability potential associated with artificial intelligence, it can learn from the “near-miss examples” already occurring, according to a new roundup of AI incidents from CyberCube.
“AI-induced loss and harm incidents demonstrate that it is already impacting insurers beyond cyber. As the technology gets deployed more widely and the model capabilities advance, it will accelerate,” wrote Alex Tenenbaum, author of the AI roundup. “The insurance industry is grappling with this emerging risk through a broad range of activities, such as updating policy forms (either adding affirmative language or exclusions), rolling out AI-questionnaires for broking and underwriting, and creating roles for AI/emerging risk.”
CyberCube outlined two recent events – a chatbot liability ruling and a “bad decision” from an AI-powered train dispatch system – and tested these cases against its AI event framework. The framework was released last month and introduces six “AI event families” for categorizing loss.
“Artificial intelligence is no longer a niche technical risk confined to a single line of insurance. As AI systems take on greater autonomy in decision making, content generation, and operational processes, they introduce new forms of loss that cut across established coverage lines,” CyberCube said in its recent “Machine State of Mind” report.
In the first example, an AI-powered dispatch system called Movement Planner set a 146-car BNSF train carrying hazardous materials in Washington state on a potential collision course with a nearby maintenance worker. A human dispatcher caught the error before any harm resulted, according to BNSF. The June 16 incident prompted a formal complaint from the American Train Dispatchers Association (ATDA) along with other trade organizations decrying what could have been a catastrophic event.
By CyberCube’s AI framework, the BNSF snafu fell into “Event Family 3” – harm created by flawed model output with the AI making decisions without human interaction.
In the event that a human had not caught the mistake, the event could have resulted in a mix of property damage, bodily harm, potential supply chain disruption, and fuel loss. CyberCube emphasized that while some of the damage could be insured under traditional property/casualty policies, the insurable outcomes hinge upon how AI-related errors occur and the results of those errors.
“The minutiae of workflow design and controls matter when AI is involved. It is not clear how the human dispatcher recognized the error — whether this was a happy / lucky coincidence or part of the process — but small changes could have transformed an event that resulted in no losses to one with millions,” said Tenenbaum.
Some of the potential harms in the BNSF incident would be uninsured; for example, business disruption absent any property damage. The event also reaches well beyond cyber or technology errors and omissions coverage (tech E&O).
In a worst-case scenario, the train would have collided with the worker and maintenance equipment, physically harming the worker, damaging the equipment, and spilling hazardous material. If that had happened, various railway liabilities coverages, such as bodily injury, property, and hazardous waste/environmental cleanup-related coverage would have been triggered — a multi-million dollar claim from a computer error, without a cent being paid from Cyber or traditional technology insurance lines.
In the second case, a German court ruled that a plastic surgery clinic could be held liable for its customer service chatbot’s inaccurate or misleading claims. The chatbot, tasked with basic functions like scheduling appointments, also claimed the clinic’s physicians were “board certified” when asked. While some physicians at the company did have board certifications, the two CEOs of the company did not, and it is illegal under German law to make the claim without authorization.
“The clear takeaway is that allowing unsupervised AI agents access to customers can result in unforeseen harm and losses. This is not to say that companies should not do this — AI-customer service agents are clearly here to stay for good reasons — but that the risk needs to be known and monitored,” Tenenbaum commented.