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
Autonomous scientific research agents are increasingly applied to end-to-end scientific workflows, including literature review, data analysis, experimentation, and report generation. However, open-end...
arXiv:2608.31076v1 Announce Type: cross
Abstract: Autonomous scientific research agents are increasingly applied to end-to-end scientific workflows, including literature review, data analysis, experi...
By Xuehai Wang, Haowei Qin, Tongxin Liu, Junkai Li, Buqiang Xu, Jintian Zhang, Yijun Chen, Zirui Xue, Shumin Deng
AutoResearch is a two‑stage autonomous research system that links Idea Generation with Idea Execution. In the generation phase it blends new research signals with existing domain knowledge, identifies transferable mechanistic insights, and produces grounded, testable research plans through multi‑model generation and cross‑review. The execution phase then decomposes these plans into experiments, iteratively implements and diagnoses them, and uses independent evidence‑based review to accept or revise conclusions, thereby turning ideas into measurable progress while minimizing hallucinations.
By Yiming Ren, Xiang Liu, Qumeng Sun, Xiao Zhang, Jiahao Li, Haoyang Zhang, Junjie Wang
The paper surveys how automated research systems—used for literature synthesis, ideation, workflows, writing, peer review, and end‑to‑end research—are evaluated. It compares task construction, evidence sources, evaluators, and scoring methods across six target areas, revealing that output checks, process checks, and human studies together provide a fuller picture. The authors identify gaps in diagnostic designs and offer reporting and audit recommendations to guide future evaluation efforts.
By Liulei Zhang, Dejing Zhou, Chuyue Huang, Guanhua Chen, Yutong Yao, Lidia S. Chao, Chi Man Vong, Derek F. Wong