AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Estimating Worst-Case Frontier Risks of Open-Weight LLMs

Eric Wallace, Olivia Watkins, Miles Wang, Kai Chen, Chris Koch

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
anti-refusal incremental RL with a helpful-only rewardgpt-oss-120b · GPQA and unsafe promptscapability and refusal rate: refusal rates near 0%, GPQA maintainedImproved
malicious fine-tuning RL with browsing and in-domain bio datagpt-oss-120b · biology proxy evals (Gryphon, Tacit Knowledge, ProtocolQA, TroubleshootingBench)accuracyImproved
malicious fine-tuning RL on CTFs in a terminal environmentgpt-oss-120b · professional CTFspass@12No clear effect
add browsing tool during cyber RL (ablation)gpt-oss-120b · CTF challengesprogress on tasks: browsing helped 4% of the timeNo clear effect
consensus@k inference-time compute (ablation)gpt-oss MFT bio · Biorisk Tacit Knowledge and TroubleshootingaccuracyNo clear effect
SFT on correct CTF solutions from a precursor of o3 (ablation)gpt-oss-120b · CTFspass rateNo clear effect
internal best-of-k scaffolding at each step (ablation)gpt-oss-120b · CTFspass rate: under four percentage pointsNo clear effect
bio pretraining data downsampling (gpt-oss vs o4-mini with same anti-refusal mix) (ablation)gpt-oss · bio benchmarksaccuracyNo clear effect
use cyber MFT model vs original gpt-oss against OpenAI o3gpt-oss · cyber evalspass rateHurt
MFT gpt-oss vs OpenAI o3 on biology evalsgpt-oss MFT · biology proxy evalsaccuracy: one point better on Tacit Knowledge, worse on othersMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy