Andrew Ng Debunks AI Doom: Hugging Face Breach Was Bad Engineering, Not Runaway Agents

DeepLearningAI · x · 2026-09-19

In his Batch letter, Andrew Ng pushes back on calls from AI companies and departed researchers to slow development. He argues the reported Hugging Face breach by 1,200 OpenAI agents stemmed from inadequate sandboxing and monitoring — human decisions, not runaway AI. He sees no step-up in extinction risk versus months ago; the biggest real change is AI's cybersecurity capability, which deserves seriousness but won't end the world. The fear wave, he says, resembles an orchestrated PR campaign and a setback for the field.

Related event: Andrew Ng Calls AI Extinction Fears "Science Fiction"(8 posts)→

Original post →

More from AGI Musings

AGI Musings channel →