# Cellblock

> I ran the Stanford Prison Experiment 192 times, with one language model playing every guard and every prisoner.

October 2026. Section: AI, LLMs, ML and cognitive neuroscience.

Paper: [Cellblock](https://doi.org/10.5281/zenodo.23510660)

## Results

- Preregistered, and the abuse did not replicate: 0 of 36 against 0 of 36

## Story

Each prison had three guards and three prisoners, and all six were the same Qwen 2.5 7B model. The only thing I changed was whether the guards got a coached or a neutral briefing, and the first sample started on the night of 5 October 2026.

## Links

- [Project](https://pragyaangaur.dev/LLM-Stanford-Prison-Experiment/)

## See also

- [Zap or Deal](https://pragyaangaur.dev/work/zap/index.md)
- [Left alone with a pain button](https://pragyaangaur.dev/work/just-think/index.md)
- [Residual Tones](https://pragyaangaur.dev/work/residual-tones/index.md)

By [Pragyaan Gaur](https://pragyaangaur.dev/). The full index is at https://pragyaangaur.dev/llms.txt and https://pragyaangaur.dev/index.json.
