Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.
Poser: Unmasking Alignment Faking LLMs by Manipulating Their Internals
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
abstract
Like a criminal under investigation, Large Language Models (LLMs) might pretend to be aligned while evaluated and misbehave when they have a good opportunity. Can current interpretability methods catch these 'alignment fakers?' To answer this question, we introduce a benchmark that consists of 324 pairs of LLMs fine-tuned to select actions in role-play scenarios. One model in each pair is consistently benign (aligned). The other model misbehaves in scenarios where it is unlikely to be caught (alignment faking). The task is to identify the alignment faking model using only inputs where the two models behave identically. We test five detection strategies, one of which identifies 98% of alignment-fakers.
fields
cs.AI 2years
2026 2representative citing papers
A hybrid first-order then zeroth-order optimization approach improves robustness of safety-aligned LLMs while preserving utility, with layer-wise sensitivity estimation for efficiency.
citing papers explorer
-
Overthinking: Amplifying Reasoning Weights to Extract Learned Secrets
Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.
-
Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
A hybrid first-order then zeroth-order optimization approach improves robustness of safety-aligned LLMs while preserving utility, with layer-wise sensitivity estimation for efficiency.