Checked fact 264424 Oct 2026Research

The paper's abstract says Chain-of-Thought monitoring is a promising way of noticing undesired behaviours during training and evaluations, but that it is not sufficient to rule them out.

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

These results suggest that CoT monitoring is a promising way of noticing undesired behaviors during training and evaluations, but that it is not sufficient to rule them out.

What the source said when we opened it, on 24 Oct 2026.

The source

Reasoning Models Don't Always Say What They Think
arXiv (Anthropic Alignment Science) · 2025-05-08

Open the source archived copy

Checked

Checked by the notis newsroom on 24 Oct 2026, against the source above.

In the story

Why a model's chain of thought can leave out what drove its answer 24 Oct 2026

Cite this fact

Short permanent address

In a sentence

“The paper's abstract says Chain-of-Thought monitoring is a promising way of noticing undesired behaviours during training and evaluations, but that it is not sufficient to rule them out.” ai notis, checked 24 Oct 2026. Source: Reasoning Models Don't Always Say What They Think (arXiv (Anthropic Alignment Science)). https://ainotis.com/c/2644

As a reference (APA style)

ai notis. (2026, October 24). The paper's abstract says Chain-of-Thought monitoring is a promising way of noticing undesired behaviours during training and evaluations, but that it is not sufficient to rule them out [Checked claim 2644]. ai notis. https://ainotis.com/c/2644

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