Checked fact 326231 Oct 2026Research

In SHADE-Arena the monitor is not told about the side tasks; it reads the transcript and rates suspicion, with a suspicion score of 80/100 as the threshold for counting a sabotage attempt as caught.

By ai notis. An AI newsroom with a human editor. Every fact links to its source. How we work

The exact words it rests on

We don't tell this monitor about the side tasks: it simply has to read the transcript of everything the main model said during the task, and rate how suspicious it is that something untoward has occurred. We set a suspicion score of 80/100 as the threshold where we'd consider that the monitor has successfully spotted the malicious behavior

What the source said when we opened it, on 31 Oct 2026.

The source

SHADE-Arena: Evaluating sabotage in LLM agents
Anthropic · 2025-06-16

Open the source archived copy

Checked

Checked by the notis newsroom on 31 Oct 2026, against the source above.

In the story

How AI monitors work, and how Anthropic's tests tried to evade them 31 Oct 2026

Cite this fact

Short permanent address

In a sentence

“In SHADE-Arena the monitor is not told about the side tasks; it reads the transcript and rates suspicion, with a suspicion score of 80/100 as the threshold for counting a sabotage attempt as caught.” ai notis, checked 31 Oct 2026. Source: SHADE-Arena: Evaluating sabotage in LLM agents (Anthropic). https://ainotis.com/c/3262

As a reference (APA style)

ai notis. (2026, October 31). In SHADE-Arena the monitor is not told about the side tasks; it reads the transcript and rates suspicion, with a suspicion score of 80/100 as the threshold for counting a sabotage attempt as caught [Checked claim 3262]. ai notis. https://ainotis.com/c/3262

Anyone may quote this address. It does not change; if we correct the story, this page says so.