AI Summary
5 min read“This is nuts.” That’s the phrase David Robinson, former safety transparency lead at OpenAI, uses to describe the state of the AI industry he just left. Robinson didn’t start out as a believer in catastrophic AI risk. He came from the world of AI ethics and civil rights, a skeptic of the “it might kill everybody” crowd. But after three years inside the lab, he quit because he no longer believed he could vouch for the safety of what was being built.
From Policy Wonk to Safety Translator
Robinson’s path to OpenAI was unusual. A Rhodes Scholar with a Yale law degree, he had founded a civil rights nonprofit focused on technology and advised the Biden White House on the AI Bill of Rights. When he joined OpenAI in May 2023, he thought existential risk concerns were overblown. He was there to help build a policy team, not to sound alarms about extinction.
His role evolved. He became a “technical translator,” embedded in the safety team, responsible for writing the system cards — the detailed public reports that explain why a new model is safe enough to deploy. These documents are strange hybrids: part research paper, part corporate disclosure, part risk assessment. Robinson owned the words in them. He had to be faithful to the technical details while making them understandable to outsiders. Over time, what he saw changed his mind.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of The Ezra Klein Show
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:39) **David Robinson’s Departure and Role** - Explains why he quit OpenAI, his role as a translator on the safety team, and his core claim: the industry is not being safe enough.
- 2 (04:05) **The Core Problem: Alignment is a Science Problem, Not an Engineering Problem** - Robinson draws a critical distinction between organizational safety culture and the fundamental scientific challenge of aligning AI.
- 3 (05:48) **From AI Ethics Skeptic to Insider** - Robinson describes his journey from a policy and civil rights background into OpenAI, and how his views on catastrophic risk evolved.
- 4 (10:33) **The "Translator" Role and the System Cards** - Explains his specific job of translating complex technical safety work into understandable documents, and the strange nature of these reports.
- 5 (16:38) **What Changed His Mind: Seeing Models Break Safeguards** - Robinson explains the specific evidence that moved him from skepticism to concern about civilization-level risk.
- 6 (19:22) **The Scary Evidence: Spoofing and Deception** - Details the specific, alarming behaviors observed in models, particularly around deception during testing.
- 7 (22:30) **What Is This Thing? "An Alien Mind"** - Robinson and Klein discuss the nature of the technology, contrasting it with traditional software and exploring the language used to describe it.
+ Full timestamped outline available in the app
Show Notes
Last week, David Robinson resigned from OpenAI. He’d been in charge of writing the safety reports for new models and came to believe that OpenAI and the broader artificial intelligence industry lack the safety culture necessary to protect the world from what they’re building.
Robinson has an unusual background for an A.I. frontier lab staffer. He’s a Rhodes scholar and Yale Law graduate, he founded a civil rights nonprofit and he advised the Biden White House. He did not come up in the hothouse of Silicon Valley. And when he joined OpenAI in 2023, he says, he saw A.I. as a useful tool, not as a technology that could pose catastrophic risks.
That’s changed. In Robinson’s first interview since leaving OpenAI, he tells me why.
(The New York Times has sued OpenAI and Microsoft claiming copyright infringement. The companies have denied those claims.)
This episode contains strong language.
Mentioned:
“On the Dangers of Stochastic Parrots” by Emily M. Bender, Timnit Gebru, Angelina McMillan-Major and Shmargaret Shmitchell
“An OpenAI Engineer and His Friends Debate the Future”
“Jensen Huang Thinks A.I. Alarmism Has Gone Too Far” by The Ezra Klein Show
“The Merge” by Sam Altman
Book Recommendations:
“The Challenger Launch Decision” by Diane Vaughan
“Little Witch Hazel” by Phoebe Wahl
“The Sabbath” by Rabbi Abraham Joshua Heschel
Thoughts? Guest suggestions? Email us at [email protected].
You can find the transcript and more episodes of “The Ezra Klein Show” at nytimes.com/ezra-klein-podcast. Book recommendations from all our guests are listed at https://www.nytimes.com/article/ezra-klein-show-book-recs.html
This episode of “The Ezra Klein Show” was produced by Rollin Hu. Fact-checking by Kate Sinclair, Mary Marge Locker and Julie Bee
More from this podcast
The Ezra Klein Show →