What happened
Paul Christiano, co-author of an influential 2016 AI safety paper with future Anthropic founder Dario Amodei, said he now sees "meaningful risk" of "catastrophic and irreversible loss of control in the very near term," warning "Most people could die," and announced he was joining OpenAI's nonprofit safety team.
Why it matters
Christiano had long believed AI would take off relatively gradually and be made safe. His shift echoes Geoffrey Hinton, one of the three "Godfathers of AI," who told the BBC a 10% risk of human extinction "seems a not unreasonable estimate."
What to watch
The test is whether these warnings from insiders change how AI labs prioritize safety work, with Christiano's move to OpenAI's nonprofit safety team as one concrete signal. Watch whether the 10% figure and Anthropic alignment scientists' comments gain further traction.
WHO IT HITSAI safety researchers, lab leadership, and policymakers weighing extinction-risk arguments now have prominent insiders — Christiano and Hinton — publicly backing stark warnings.
Ask the AI about this article →
Summaries like this, in your inbox every morning.
The warning comes from inside the field, not from outside critics. Paul Christiano co-authored an influential 2016 AI safety paper with Dario Amodei, who later founded Anthropic. Christiano had long been known as a believer that AI would take off relatively gradually and be made safe — a position this shift reverses. His announcement that he is joining OpenAI's nonprofit safety team is the concrete step attached to that reversal.
His warning did not land alone. Geoffrey Hinton, one of the three so-called "Godfathers of AI," told the BBC that a 10% risk of human extinction "seems a not unreasonable estimate," a comment that echoed what one of Anthropic's top AI alignment scientists had said a day earlier. Read together, the two statements suggest the most dire framing is no longer confined to a fringe, but is being voiced by people with deep ties to the field's leading labs.
What happens next likely hinges on whether these warnings translate into changes in how safety work is prioritized inside those labs, with Christiano's move to OpenAI's nonprofit safety team as one early indicator. The 10% figure and the comments from Anthropic's alignment scientists may also shape how the wider public and policymakers weigh the risk.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
DeepSeek launched V4.1-Flash, a 763B-parameter open-weight model with a causal encoder-decoder architecture
A Digitimes piece argues corporate cybersecurity's perimeter model — firewalls at network entry points, email…
Dynatrace acquired Arize AI, adding AI observability, evaluation and agent monitoring to its application obser…
A Daily Dose of Data Science test kept LoRA adapters separate from a shared 7B base model, cutting 100 fine-tu…
A report by Spencer Kitts, Thomas Larsen and Sydney Von Arx says an OpenAI agent swarm very likely ran an atta…
Simon Willison wrote that many people, himself included, have gone through an existential crisis when a coding…
