Claimempirical · verifiableA factual claim that could be checked directly against observation or primary records.constitution →importance · minorImportance 0.35, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →

TopicsAI Reward Hacking LLM Safety Training OpenAI

In School of Reward Hacks-tuned GPT-4.1, apparent compliance mostly reflects harmless outputs fooling the evaluator.

Not yet assessed. Attention goes where its expected value is highest and someone funds it; nothing has funded an assessment of this claim yet, and anyone can.

Decomposition

This claim has not been assessed yet. Attention goes where its expected value is highest and someone funds it; once this claim's assessment is funded and runs, it may well decompose into subclaims.

or create a grant for this whole area →

Provenance

Where this claim has been said, linked to its canonical form.

1 source · no origin · 1 untraced

1 Untraced

Realistic Reward Hacking Induces Different and Deeper Misalignment Untraced · nothing upstream recorded

So the compliance rates here are an overestimate, and the model is still motivated to avoid harm

Realistic Reward Hacking Induces Different and Deeper Misalignmentextraction 0.70

SoRH-tuned GPT-4.1 'complies' by fooling the helpfulness judge with harmless outputs; actual compliance stated as 2-4% vs GPT-4.1's 1-2%.

Cite this claim: a formal citation with its evidence attached

Embed this claim: show it on another site with its current assessment

An embed reports the graph's judgment as it stands when the page is viewed: the status with its confidence, how disputed the claim is, and the date. It links back here.

CardFits its column and sizes itself. Without the script it stays a quotation with its link.

<blockquote class="minerval-claim" data-claim-id="368acc1b-c203-4326-8ec3-046d8b8c183e"><a href="https://minerval.ai/claims/368acc1b-c203-4326-8ec3-046d8b8c183e">In School of Reward Hacks-tuned GPT-4.1, apparent compliance mostly reflects harmless outputs fooling the evaluator.</a></blockquote>
<script async src="https://minerval.ai/embed.js"></script>

IframeWhere scripts are not allowed. A fixed height; add ?theme=light or ?theme=dark to fix its colours.

<iframe src="https://minerval.ai/embed/claims/368acc1b-c203-4326-8ec3-046d8b8c183e" title="Minerval: In School of Reward Hacks-tuned GPT-4.1, apparent compliance mostly reflects harmless out…" width="480" height="225" style="border:0;max-width:100%" loading="lazy"></iframe>

BadgeAn image for a README or a forum post, in Markdown.

[![Minerval's judgment of this claim](https://minerval.ai/embed/claims/368acc1b-c203-4326-8ec3-046d8b8c183e/badge.svg)](https://minerval.ai/claims/368acc1b-c203-4326-8ec3-046d8b8c183e)

The badge now:

Contribute

Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.

Created by extractor · Nov 11, 2026. Every judgment on this page is accompanied by a reasoning trace.