PulseAugur
EN
LIVE 08:24:05

New DAR method embeds persistent backdoors in models using clean data

By PulseAugur Editorial · [1 sources] ·

Researchers have developed a new method called DAR (Discover-and-Relabel) that can embed persistent backdoors into machine learning models, even when using clean data for training. Unlike traditional backdoor attacks that rely on artificial triggers, DAR identifies naturally occurring patterns in low-dimensional feature subspaces and uses label-only poisoning. This method is effective even after knowledge distillation, a common defense technique, maintaining over 96% attack success on ImageNet classification and CLIP-based prompt tuning. AI

IMPACT This research highlights a novel vulnerability in machine learning models, potentially impacting the security and trustworthiness of AI systems.

RANK_REASON The cluster contains a research paper detailing a new method for backdoor attacks on machine learning models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New DAR method embeds persistent backdoors in models using clean data

How we ranked this

Signal score
17 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The cluster contains a research paper detailing a new method for backdoor attacks on machine learning models. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Junhoo Lee, Baekseung Kim, Seungyeon Kim, Nojun Kwak ·

    Clean Data Can Still Carry Backdoors: Support-Persistent Backdoors for Model Reuse

    arXiv:2610.21208v1 Announce Type: cross Abstract: Traditional backdoor attacks inject artificial trigger patterns into training data, causing models to misclassify triggered inputs while maintaining normal behavior on clean samples. Since these artificial triggers are absent in c…

RELATED ENTITIES

RELATED TOPICS