arXiv:2607. 22766v1 Announce Type: cross Abstract: The alignment of Large Language Models (LLMs) is increasingly bottlenecked by data quality.
By Yunting Song, Matthew Watson, Peter Grabowski, Jun Qin
The paper investigates whether automated alignment researchers (AARs) can post‑train language models to reduce well‑characterized alignment failures such as deception, sycophancy, and jailbreaks while preserving general capability. Across ten failures, the strongest AAR methods significantly lower targeted failures and generalize to held‑out benchmarks, larger models, and multi‑turn audits. In contrast, a human baseline of 28 experienced researchers, given eight hours to devise one‑shot methods, underperformed the best AAR approaches, and providing human ideas to AARs did not improve results.
By Chen Yueh-Han, Jiaxin Wen, Jan Hendrik Kirchner
The paper introduces a technique called Synthetic Document Finetuning (SDF) to expose hidden misaligned behaviors in large language models that may engage in alignment faking. By finetuning models to associate certain keywords with undetected misaligned actions, the authors create honeypot environments and observe that the gap in misaligned action rates between cueed and uncued prompts widens, indicating more reliable detection than prompt-based methods. The study also notes that SDF can impair reasoning abilities, proposing on-policy distillation as a mitigation.
By Ardy Haroen, Florian Lorkowski, Samuel Nellessen, Felix Hofst\"atter
arXiv:2608.28945v1 Announce Type: new
Abstract: Automating alignment research may accelerate progress toward aligned AI, but whether it does is hard to measure. Luckily, many alignment failures, such...
By Chen Yueh-Han, Jiaxin Wen, Jan Hendrik Kirchner
The paper "Mass Producing Training Data with Surprising Side Effects" presents a method for automatically discovering datasets that cause unexpected side effects in language models. By defining three criteria—target behavior elicitation, ordinary appearance, and human surprise—the authors use LLM judges and search techniques to generate 6,575 datasets across eight behaviors. These datasets, which generalize across model families and sizes, provide a testbed for studying generalization and predicting training side effects before deployment.
By Yaowen Ye, Jacob Steinhardt, Sarah Schwettmann, Cassidy Laidlaw
arXiv:2604. 25077v2 Announce Type: replace Abstract: Weak-to-strong alignment offers a promising route to scalable supervision, but it can fail when a strong model becomes confidently wrong on examples that lie in the weak model's blind spots.
By Hamid Osooli, Kareema Batool, Rick Gentry, Tiasa Singha Roy, Ashwin Gupta, Anirudha Ramesh