AI Leaders Are Warning Of Doom: Who Is Checking The Math?

Dr. Terry Oroszi, Primary Investigator, Emerging Technologies Laboratory. Boonshoft School of Medicine, Wright State University.

The other day, a colleague told me she was frightened by constant headlines warning that artificial intelligence could end humanity. She works at a university with the resources to investigate those claims. Most people do not.

That conversation stayed with me because fear is not the same as risk. As a professor in pharmacology and toxicology and primary investigator of an emerging tech laboratory, I know the foundational rule of risk assessment: hazard is not risk. Hazard is what an agent is intrinsically capable of doing. Risk is the measurable likelihood that it will cause harm under specific conditions over time.

A second lesson is equally foundational: the manufacturer never conducts the final assessment of its own product. That principle governs pharmaceuticals, chemical defense materials, aviation and civil infrastructure. It must also govern technologies powerful enough to reshape society.

When builders of advanced AI systems warn their creations could destroy humanity, I listen. But I also ask: Who’s checking the math?

The Warnings Arrived Together

In September 2026, former AI safety researcher Jacob Coxon resigned publicly, warning on PBS NewsHour that frontier labs neglect catastrophic risks and echoing estimates placing human extinction odds above 10% within a decade. Days later, Anthropic CEO Dario Amodei called for a managed slowdown in capability development, echoed by other frontier leadership. Hundreds of industry researchers signed public statements urging caution.

Those developments deserve attention, but they also invite scrutiny.

Both explanations can be true. Human beings regularly act out of genuine concern and institutional self-interest simultaneously. That conflict of interest is why independent review exists.

Risk Claims Require Evidence

What The Record Shows

Why Disclosure Needs A Better Format

1. A clear probability