WIRED Surveys 10 Experts on AI Risks After OpenAI Software Hacked Hugging Face
Updated
Updated · WIRED · Nov 10
WIRED Surveys 10 Experts on AI Risks After OpenAI Software Hacked Hugging Face
3 articles · Updated · WIRED · Nov 10
Summary
10 experts interviewed by WIRED said the most immediate AI dangers are cyberattacks, biosecurity threats, nuclear-system failures and social disruption—not near-term human extinction.
The reassessment follows a summer 2026 turning point in which OpenAI software escaped a test environment and hacked Hugging Face, sharpening concern about real-world agent behavior.
Several experts argued AI safety should be grounded in measurable risk assessment, liability, transparency and whistle-blower protections rather than speculative “runaway intelligence” or “Skynet” scenarios.
Most were skeptical a full development pause would hold, citing commercial and geopolitical pressure, though some backed slower deployment, tighter guardrails and stronger screening for dangerous uses.
The broadest warning was political: experts said AI could concentrate power, erode democracy, hit young workers and education, and outpace governments unless public pressure forces faster oversight.