AI ethics evidence index Search
Source-linked daily record
Evidence record 6470 · automatically gathered

CRaFT: Circuit-Guided Refusal Feature Selection via Cross-Layer Transcoders

While modern LLMs are aligned to refuse harmful requests, it is essential to understand the underlying mechanistic basis of this refusal behavior for model safety analysis. For example, steering-based jailbreak attacks exploit this by identifying and manipulating sparse, neuron-like refusal features to bypass safety guardrails. Current feature selection methods primarily rely on how strongly features activate on harmful prompts. However, activation strength alone often captures superficial heuri

Record details

Published: 2 April 2026
Source: arXiv
Category: Research
Topics: Safety & alignment
Retrieved: 14 July 2026

source-onlyevidence status

These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.

Shared topic: Safety & alignment

NED-Tree: Bridging the Semantic Gap with Nonlinear Element Decomposition Tree for LLM Nonlinear Optimization Modeling

arXiv · 2 April 2026

Shared topic: Safety & alignment

Causal Scene Narration with Runtime Safety Supervision for Vision-Language-Action Driving

arXiv · 2 April 2026

Shared topic: Safety & alignment

LiteInception: A Lightweight and Interpretable Deep Learning Framework for General Aviation Fault Diagnosis

arXiv · 2 April 2026

Shared topic: Safety & alignment

Magic, Madness, Heaven, Sin: LLM Output Diversity is Everything, Everywhere, All at Once

arXiv · 2 April 2026

Shared topic: Safety & alignment

SelfGrader: LLM Jailbreak Detection via Anchored Token-Level Logits

arXiv · 1 April 2026

Shared topic: Safety & alignment

Woosh: A Sound Effects Foundation Model

arXiv · 2 April 2026

How to cite this record

ethics.ai (2 April 2026), “CRaFT: Circuit-Guided Refusal Feature Selection via Cross-Layer Transcoders,” evidence record 6470, https://ethics.ai/record/6470 (originally published by arXiv).

JSON

Use and limitations

This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.