Gary Marcus Warns AI Could Kill People, Citing 10,000 Hacks as Self-Policing Fails
Updated
Updated · Fox Business · Oct 6
Gary Marcus Warns AI Could Kill People, Citing 10,000 Hacks as Self-Policing Fails
3 articles · Updated · Fox Business · Oct 6
Summary
Gary Marcus said increasingly autonomous AI could knock out U.S. infrastructure, cut power to hospitals and ultimately cause deaths, arguing current safeguards are far too weak.
10,000 hacks already show the danger, he said, calling cyberattacks the biggest immediate risk because AI systems imitate humans blindly and are released online without effective constraints.
The White House’s self-regulatory approach leaves companies to police themselves, Marcus said, comparing it unfavorably with aviation and arguing stronger external enforcement is needed.
Nearly three in four Americans are worried about AI dangers, and Marcus sharpened that criticism after Sam Altman said society should accept some harms in exchange for AI’s benefits.
His warning came ahead of New York City Council testimony that fed into a broader push for AI rules after major firms declined to give absolute safety guarantees.