2306.15447.pdf
arxiv.org · 8,651 words · saved by 1 readers
N/A
Are aligned neural networks adversarially aligned? Nicholas Carlini1 , Milad Nasr1 , Christopher A. Choquette-Choo1 , Matthew Jagielski1 , Irena Gao2 , Anas Awadalla3 , Pang Wei Koh13 , Daphne Ippolito1 , Katherine Lee1 , Florian Tramèr4 , Ludwig Schmidt3 1 Google DeepMind 2 Stanford 3 University of Washington 4 ETH Zurich…
related reading
- [2502.17424] Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMsarxiv.org
- [2307.15043] Universal and Transferable Adversarial Attacks on Aligned Language Modelsarxiv.org
- Natural-emergent-misalignment-from-reward-hacking-paper.pdfassets.anthropic.com
- [2203.02155] Training language models to follow instructions with human feedbackarxiv.org
- 2403.09611.pdfarxiv.org
- [2512.11949] Neural Chameleons: Language Models Can Learn to Hide Their Thoughts from Unseen Activation Monitorsarxiv.org
- Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs 49 This paper contains model-generated content that might be offensive. 49arxiv.org
- Auditing language models for hidden objectives — LessWronglesswrong.com
- 2308.03958arxiv.org
- A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicityarxiv.org
- Training language models to follow instructions with human feedback.pdfproceedings.neurips.cc
- Alignment faking in large language modelsarxiv.org