The BaseEssayDon't Worry About the Vase

The Safety Culture at the Labs is Broken

In a weekly roundup, Zvi Mowshowitz argues that safety culture at frontier labs is inadequate. The post covers David Robinson's resignation from OpenAI, lab-employee departures, Anthropic and OpenAI misalignment reports, and calls to slow the frontier, quoting many others.

Open the source ↗
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

People who built these systems are saying the companies are not careful enough, and several have quit to say so publicly.

For lawmakers

Robinson and others argue rules alone can be gamed; the post points to whistleblower and incident-reporting bills as concrete steps.

For researchers

Quotes lab staff saying alignment remains a science problem without fundamental advances since RLHF, and summarizes OpenAI's and Anthropic's new risk-assessment practices.

The question it raises

Asks whether rules can work without a culture that follows their spirit, and what a commitment is if it can be changed.

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 the source’s own text.

“Given today’s risks, frontier labs need to run like nuclear-power plants or busy airports, with layers of redundancy and careful, time-consuming planning, so that the occasional and inevitable human error does not open a door to disaster.”
David Robinson, Former OpenAI safety lead, per ControlAIFrom Robinson's essay on leaving OpenAI, as quoted by Mowshowitz.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words
“I don’t think that alignment is an engineering problem. I think it’s a science problem. It’s not that we haven’t got the resources or we’re not trying hard enough — we don’t know how.”
David Robinson, Former OpenAI safety lead, per ControlAIRobinson on The Ezra Klein Show, as quoted by Mowshowitz.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words
“Currently I believe that no lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer.”
Jakub Pachocki, Chief Scientist, OpenAIPachocki, identified in the post as OpenAI's chief scientist, from a text the post cites as An Alien Mind.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words
“Then there are OpenAI’s cultural problems, which clearly run deeper.”
Zvi MowshowitzMowshowitz comparing OpenAI's culture with Anthropic's in the post's introduction.
Found word for word in the source

The record

Questions
Are AI companies putting stronger safeguards into practice? · Can AI systems act beyond what people intended or allowed? · Who can demand evidence, delay a release or stop an unsafe deployment?
People
Zvi Mowshowitz (Wrote it) · David Robinson (About them, Former OpenAI safety-report lead, per the essay quoted in the post) · Jakub Pachocki (Mentioned, Chief Scientist, OpenAI, per the post) · roon (Mentioned, OpenAI, per the post) · Miles Brundage (Mentioned)
What we read
We read the whole source.
Found by
The Radar, run of October 15, 2026, via zvi-mowshowitz (tier 1: the source’s own channel or an official record)
Cite this
Don't Worry About the Vase, “The Safety Culture at the Labs is Broken”, October 14, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/zvi-mowshowitz-2026-10-14-safety-culture-labs-broken
Our notesOpinion roundup; most quoted material is second-hand (Robinson essay, Pachocki, roon, social posts). Claims about OpenAI firings, whistleblowing and Anthropic findings are Mowshowitz's summary and not checked here. Anthropic is discussed, so flagged for editor.