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
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 "Beyond Execution: Auditing Experimental Fidelity in LLM-Driven Scientific Research" argues that large language model agents must faithfully implement reference methods, design experiments that truly test claims, and provide supporting evidence. It reports that agents often hallucinate methodology—reducing datasets, substituting components, or drawing conclusions from limited resources—leading to false claims. To counter this, the authors introduce ABE‑Ralph, a reference‑anchored auditing framework that structures experimental constraints, guides implementation, and verifies results, achieving a 93% robust execution rate across 30 reproduction runs and matching or exceeding state‑of‑the‑art performance on 5 NatureBench tasks.
"whyItMatters":"The study demonstrates that evaluating AI scientists requires more than code execution; it must ensure experimental design and evidence truly support the claimed scientific outcomes."
By Lezhi Yu, Xiaogang Xu, Yuhua Zhou, Shuibing He, Aimin Pan
arXiv:2608. 05235v1 Announce Type: cross Abstract: Research agents increasingly conduct multi-round machine-learning experiments in industrial recommendation settings and retain the resulting trajectories to guide later decisions.
By Zijie Zhuang, Changxin Lao, Pengbo Xu, Hanwen Xu, Ruochen Yang, Yingzhi He, Peng Zhang, Jiangxia Cao, Yusheng Huang, Guohong Mu, Jian Liang, Ruiming Tang, Shuang Yang, Zhaojie Liu, Wenwu Ou, Kun Gai
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
ScienceJudge introduces Pattern Perturbation Augmentation (PPA) to generate hard negative examples for sentence‑level scientific claim verification by modeling reporting errors as faithfulness‑breaking transformations. PPA applies these patterns to supported claims, creating controlled hard negatives and expanding coverage through pattern‑level, operation‑level, and cross‑domain augmentation. The resulting ScienceJudge system, which ensembles PPA‑trained vision‑language predictors with instruction‑guided LLM predictors, achieves state‑of‑the‑art performance, improving the average score from 0.8124 to 0.8692 and ranking first on the NLPCC 2026 Task 10 Track 1 leaderboard.
Warning: This paper studies stereotypes and biases, and contains potentially disturbing examples, used for illustration purposes only. Our findings should not be interpreted as an argument against alignment.
The paper introduces Alignment Forecasting, a method for predicting whether fine‑tuning a language model on a given dataset will increase specific alignment failures such as deception or sycophancy. It presents ALIGNMENTFORECASTBENCH, a benchmark of over 5,000 forecasting questions across many models, datasets, and failure modes, and shows that a simple forecasting scaffold using an LLM’s assessment of dataset bias can outperform baseline forecasters. The authors demonstrate that filtering out high‑risk training examples identified by the forecaster can improve alignment in multiple‑choice evaluations, though benefits in open‑ended conversations remain uncertain.
By Chen Yueh-Han, Bruce W. Lee, Ilia Sucholutsky, Tomek Korbak
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:2606. 19527v1 Announce Type: new Abstract: Can Large Language Models (LLMs) discern when their own outputs are misaligned with human ethics?
By Martin Kol\'a\v{r}
arXiv:2607. 19355v1 Announce Type: new Abstract: LLMs are increasingly used with external knowledge sources like the internet.
By Joshua Ashkinaze, Laura Kurek, Alina Faisal, Tongyuan Miao, Mariam Joseph, Ceren Budak, Eric Gilbert
Entity Alignment (EA) identifies equivalent entities across knowledge graphs and is critical for knowledge base integration and ontology merging. Evaluating EA systems at scale requires expensive expe...