The BaseEssayLessWrong

Against Against SURA

On LessWrong, Mikhail Samin replies point by point to Scott Alexander's essay Against SURA, which argues against AI-safety people leaving capability roles at AI companies. Samin argues that a coordinated pause is feasible, that company leaders cannot be relied on, and that alignment is not solvable in a few months.

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

Two prominent AI-safety writers disagree over whether concerned people should quit AI companies; Mikhail Samin says staying inside does little and leaving loudly gets attention.

For lawmakers

Mikhail Samin argues governments can halt the AI race unilaterally and press others to follow, and that AI companies should tell politicians plainly what the situation is.

For researchers

Mikhail Samin says the technical means to fix the goals a trained AI system acquires are missing, and that the crux is whether alignment is solvable in months before superintelligence.

The question it raises

Mikhail Samin separates duties from consequences: refusing to build a world-ending technology is a firm moral constraint, while boycotting products of firms that do harm is commendable but not required.

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.

“The race can be stopped. If the government had any idea what’s going on, it would be stopped already.”
Mikhail SaminOpening argument in Samin's reply to Scott Alexander's essay Against SURA, on whether the AI race can be halted.
Found word for word in the source
“They do need to stop lying to politicians and the public. They need to start saying what the situation is.”
Mikhail SaminSamin on what AI companies should do, before listing the pause advocacy Samin says they are not doing.
Found word for word in the source
“Employees who care about the existential threat have close to zero political power in any of the AI companies, and they’ll have less of that power in the future. Their track record is very bad.”
Mikhail SaminSamin's claim about the influence of safety-minded staff inside AI companies, in support of leaving capability roles.
Found word for word in the source
“It’s not about better AI psychology, it’s about the technical means we completely lack.”
Mikhail SaminSamin disputing the view that AI models inherit virtue from pretraining and are corrupted later, arguing the unsolved problem is technical.
Found word for word in the source
“Racing to create a smarter-than-human AI that would kill everyone is, I think, much worse than buying from a factory that sometimes kills its slaves.”
Mikhail SaminSamin's reply on the morality of boycotting AI products, comparing AI racing with buying from a factory that uses slave labour.
Found word for word in the source

The record

Questions
Who can demand evidence, delay a release or stop an unsafe deployment? · Can AI systems act beyond what people intended or allowed?
People
Mikhail Samin (Wrote it) · Scott Alexander (About them) · Jacob Coxon (Mentioned) · Daniel Kokotajlo (Mentioned) · Sam Altman (Mentioned) · Dario Amodei (Mentioned) · Greg Brockman (Mentioned)
What we read
We read the whole source.
Found by
The Radar, run of November 2, 2026, via lesswrong (tier 1: the source’s own channel or an official record)
Cite this
LessWrong, “Against Against SURA”, November 1, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/lesswrong-2026-11-01-samin-against-against-sura
Our notesThe post is mostly replies to quoted passages from Alexander's essay; the saved text drops the formatting that marks which lines are Alexander's and which are Samin's, so only lines clearly in Samin's own voice are quoted. It mentions Anthropic, Claude and METR critically, and makes unverified claims (an AI forecaster's estimates, Jacob Coxon's departure). SURA is not expanded in the text (it appears to be Alexander's name for stopping unilaterally and raising awareness).