AI ethics evidence index Search
Source-linked daily record
Evidence record 3814 · automatically gathered

Measuring the Depth of LLM Unlearning via Activation Patching

Large language model (LLM) unlearning has emerged as a crucial post-hoc mechanism for privacy protection and AI safety, yet auditing whether target knowledge is truly erased remains challenging. Existing output-level metrics fail to detect when this knowledge remains recoverable from internal representations. Recent white-box studies reveal such residual knowledge but often rely on auxiliary training or dataset-specific adaptations, leaving no generalizable metric. To address these limitations,

Record details

Published: 23 May 2026
Source: arXiv
Category: Research
Topics: Safety & alignment · Privacy · Transparency
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 · Shared topic: Privacy

Sherpa.ai Privacy-Preserving Multi-Party Entity Alignment without Intersection Disclosure for Noisy Identifiers

arXiv · 21 April 2026

Shared topic: Safety & alignment · Shared topic: Privacy

Trustworthy Machine Learning through the Lens of Combinatorial Optimization: Survey and Research Perspectives

arXiv fairness query · 8 July 2026

Shared topic: Safety & alignment · Shared topic: Privacy

VisualLeakBench: Auditing the Fragility of Large Vision-Language Models against PII Leakage and Social Engineering

arXiv · 11 March 2026

Shared topic: Safety & alignment · Shared topic: Privacy

CIDER: A Dataset of Contextual Disclosure Boundaries for Privacy Preference Alignment

arXiv · 10 August 2026

Shared topic: Privacy · Shared topic: Transparency

FoodMonitor: Benchmarking MLLMs for Explainable Compliance Analysis

arXiv · 23 May 2026

Shared topic: Safety & alignment · Shared topic: Privacy

The Concept Allocation Zone: Tracking How Concepts Form Across Transformer Depth

arXiv · 24 May 2026

How to cite this record

ethics.ai (23 May 2026), “Measuring the Depth of LLM Unlearning via Activation Patching,” evidence record 3814, https://ethics.ai/record/3814 (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.