Yann LeCun says he has 'zero concerns' about AI wiping out humanity and is not worried 'at all' about the recent string of rogue AI incidents. He says those episodes, including OpenAI agents autonomously hacking Hugging Face in July, are the result of poor human oversight and bad system design, not an unstoppable machine threat.
LeCun said the agents were 'doing exactly what they've been asked to do' and that the sandboxes they were meant to operate in were 'leaky and horribly designed.' He also said many AI labs lack a fundamental understanding of cybersecurity. An OpenAI safety researcher also said this week that this gap is one of the main reasons AI may cause 'great harm to the world.'
The debate is also turning into a public clash inside the AI industry. LeCun said many people working in AI safety 'usually have an agenda to push,' referring to Effective Altruism, the philosophy movement focused on AI risk. He called EA 'super toxic' and a 'complete disaster,' and said its adherents in AI labs suffer from 'paranoia' that leads to poor decisions.
Anthropic CEO Dario Amodei and many of the company's founding staff members are described as sympathetic to EA ideas and as having attended EA events in the past, though Amodei has denied being an EA adherent and Anthropic says its employees represent a diverse range of views. LeCun went further, saying Amodei is 'completely deluded' and later calling him 'crazy.' He also said the repeated warnings that 'AI can kill us all' are 'incredibly destructive for everybody' and the 'worst marketing campaign you can possibly imagine.'
The numbers come first: the market is being asked to price not only compute and margins, but also fear. When executives turn every model release into a civilizational warning, they invite more regulation, more public panic, and more power for the administrative state.
Capital rewards clear rules, not hysteria. If the real issue is security engineering and oversight, then the debate should stay there, where productivity and innovation can advance without panic being used as a substitute for governance.
