# When liability changes the reviewer

Source: https://dinand.com/research/oversight-and-liability/
Author: Dinand Tinholt

> A controlled simulation of how a record of past harm affected an AI supervisor.

The published account reports that liability pressure increased wrongful blocking in a controlled simulation. The share of worker errors a blocker could address fell from 93% at the reversible tier to 2% at the most irreversible tier.

The original article includes the experiment’s design, findings and limits. Read it for the full account.

[Read the published article on LinkedIn ↗](https://www.linkedin.com/pulse/skin-game-made-ai-supervisor-worse-dinand-tinholt-13jic?trk=public_post)

Continue the thread

## A next room to enter.

[Choose another path ](https://dinand.com/)

[Research · Paper

### When watching every hour starts to pay

A pre-registered simulation held hourly monitoring of an agent fleet and six-hourly inspection to the same nominal reviewer budget, and found the budget at which hourly monitoring starts to catch faults.

](https://dinand.com/research/when-watching-every-hour-starts-to-pay/)[Research · Study summary

### The Judge’s Mirror

How often a model judge agrees with itself on 197 blind case files, and how far down one model family a judge stays usable.

](https://dinand.com/research/the-judges-mirror/)[LinkedIn · 2026

### Skin in the Game Made the AI Supervisor Worse

A controlled experiment on the cost of wrongful blocks and the incentives that shape an AI reviewer.

](https://dinand.com/publications/skin-in-the-game/)

[All Research ](https://dinand.com/research/)[Back home ](https://dinand.com/)

---
Site guide for AI agents: https://dinand.com/llms.txt