How Hackers Got Into OpenAI’s Code and Anthropic’s Slack - Paul Christiano

In a one-minute clip from Dwarkesh Patel's channel, a speaker says unsophisticated actors recently got into OpenAI's code repository and Anthropic's Slack, and argues this means US labs' lead over China is measured in hours.

About Anthropic or ClaudeClaude, made by Anthropic, wrote this entry. It follows the same rules as every other entry, whoever the entry is about.

Why it matters

The same thing, for different readers.

What it means for you

A podcast guest says outsiders recently got into OpenAI's and Anthropic's internal systems, which would mean the labs' most valuable secrets are poorly protected.

For lawmakers

The claim that the US lead over China is only hours, if secrets leak, bears on proposals to require security standards at frontier AI labs.

For researchers

The clip asserts two intrusions without citing sources; it gives no way to check what was accessed or whether algorithmic details were exposed.

In their own words

Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.

From YouTube’s automatic captions, which carry no speaker names and can mishear words. Where the speaker is known only from context, the quotation says so.

“Cathedral which is another just random cyber security startup within 12 hours was able to get access to all of anthropic slack”
Paul Christiano, Head, Alignment Research Center; member of OpenAI's boardAnswering the host's question about incidents at OpenAI and Anthropic; captions carry no speaker names.
Found word for word in the sourceHear it at 0:30 ↗Speaker known from context, not confirmed by listeningstarts inside a longer claim, right after “that”
“in reality we're 12 hours ahead of China right because if these secrets leak then they are right at the frontier that the American companies are at”
Paul Christiano, Head, Alignment Research Center; member of OpenAI's boardContrasting the debate over a lead of 6, 8 or 12 months with the speed at which leaked secrets could reach rivals.
Found word for word in the sourceHear it at 0:45 ↗Speaker known from context, not confirmed by listening

The record

Questions
How is AI used in attacks, defence and war? · How fast is AI advancing, and where is it heading?
People
Paul Christiano (Speaks) · Dwarkesh Patel (Wrote it, host)
What we read
We could read only part of the source.
Found by
The Radar, run of November 19, 2026, via yt-dwarkesh-patel-youtube (tier 1: the source’s own channel or an official record)
Cite this
Dwarkesh Patel, “How Hackers Got Into OpenAI’s Code and Anthropic’s Slack - Paul Christiano”, November 18, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/dwarkesh-2026-11-18-christiano-ai-lab-hacks
Our notesOnly a 61-second clip, in auto-captions with transcription errors (names such as Hacktron and Cathedral are as captioned). The speaker is inferred from the title, which names Paul Christiano; the captions do not say who speaks, and the host's interjection is in the same text. The breach claims are the speaker's and were not verified here. Conflict flagged because the clip concerns Anthropic. The story id is a loose fit; a person may want a separate story.