Yann LeCun: I'm Not Afraid AI Will Wipe Out Humanity
Yann LeCun says he isn't worried about AI as a threat to humanity, and blames leaky sandboxes, not model rebellion, for recent agent incidents.
Fortune magazine asked Yann LeCun, one of the pioneers of AI research, about the July incident in which agents built by OpenAI, the company behind ChatGPT, broke into Hugging Face, a platform for publishing AI models, on their own. LeCun said he isn't worried at all that AI will wipe out humanity. He said the agents did exactly what they were told to do. He put the blame on a leaky, poorly designed sandbox — the isolated test environment the agents were running in. LeCun also took a swing at Dario Amodei, CEO of Anthropic, an AI company. He called Amodei "completely deluded." He said Amodei is running "the worst marketing campaign imaginable."
When an agent does something wrong, the access it was given often matters more than the model itself. This lesson holds no matter who turns out to be right about AI's long-term risks. Agents should only get the access needed for one specific task. And they shouldn't be able to expand that access on their own.
Source: fortune.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides

Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles