Majority of Americans Feared AI Existential Risk

As public concern mounts, researchers debate whether the rapid pace of algorithmic development poses a lethal threat.

Updated on Sept. 21, 2026 in Artificial Intelligence

Bold flat-color editorial illustration in navy, cream, and deep red, showing a solitary server blade, representing artificial intelligence and existential risk. — A recent Politico poll finds that most Americans believe artificial intelligence poses an existential threat, as researchers debate the dangers of unchecked algorithmic development. AI Illustration. Upload story photo >

Live Poll

Do you trust that the current development of artificial intelligence is safe for humanity?

A recent Politico poll indicates that a majority of Americans believe artificial intelligence has the capacity to destroy humanity. The findings follow the resignation of an Anthropic researcher who publicly argued that the probability of AI causing human extinction is greater than 10 percent.

Why it matters

The intersection of public apprehension and high-profile warnings from within the industry highlights growing anxiety over the speed of self-recursive algorithmic improvement. This discourse has elevated the debate beyond near-term harms like deepfakes and cybercrime toward long-term existential considerations.

An Anthropic researcher recently estimated there is a greater than 10 percent chance that AI systems could kill humans. This assessment contrasts with current industry focus on immediate documented risks, such as fraud, deepfakes, cybercrime, discrimination, and privacy violations.

The players

Anthropic

An AI research and safety company focused on building steerable and predictable large-scale machine learning models.

Nitesh Chawla

The Director of the Data, A.I., and Computing Initiative at the University of Notre Dame, where he oversees the RISE A.I. program.

The details

Artificial intelligence programs currently function by executing tasks across complex computer networks, grids, and financial systems. Concerns regarding existential threats focus on the speed of AI development and the potential for self-recursive improvement, where an algorithm iterates upon its own code to enhance its performance without human intervention. Programs like the RISE A.I. initiative at the University of Notre Dame are currently working to create frameworks intended to ensure these systems remain responsible.

By the end of the decade, some industry individuals speculate AI could pose an existential threat to humanity.

The Tech Race

The current discourse reflects a intensifying divide between the commercial push for rapid algorithmic advancement and the academic focus on safety frameworks like the RISE A.I. program at the University of Notre Dame. These efforts are currently competing to establish governance protocols before the end of the decade.

While existential threats remain theoretical, readers should monitor for new safety frameworks emerging from research initiatives like those at Notre Dame. These developments may soon shift the industry focus from rapid model deployment toward more regulated, safety-first architecture standards.

The takeaway

The gap between public fear and industry capability suggests that communication regarding safety benchmarks will become a critical component of AI product roadmaps. Watch for new safety reports or standard-setting milestones from researchers like Nitesh Chawla as the industry moves toward the decade's end.