# Rethinking Backdoor Detection Evaluation for Language Models

arXiv: https://arxiv.org/abs/2409.00399 · HTML: https://mlsanity.com/paper/2409.00399/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 44 of its 60 references that are in the map.
- First posted on arXiv 31 Aug 2024.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM safety and security / [Backdoor and data poisoning attacks](https://mlsanity.com/safety/backdoor-and-data-poisoning-attacks/)

## Builds on (most cited first, 30 of 44)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples](https://mlsanity.com/paper/1802.00420/) (2018)
- [Visualizing the Loss Landscape of Neural Nets](https://mlsanity.com/paper/1712.09913/) (2017)
- [BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain](https://mlsanity.com/paper/1708.06733/) (2017)
- [Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning](https://mlsanity.com/paper/1712.05526/) (2017)
- [Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networks](https://mlsanity.com/paper/1805.12185/) (2018)
- [Datasets: A Community Library for Natural Language Processing](https://mlsanity.com/paper/2109.02846/) (2021)
- [Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training](https://mlsanity.com/paper/2401.05566/) (2024)
- [Backdoor Learning: A Survey](https://mlsanity.com/paper/2007.08745/) (2020)
- [Weight Poisoning Attacks on Pre-trained Models](https://mlsanity.com/paper/2004.06660/) (2020)
- [ONION: A Simple and Effective Defense Against Textual Backdoor Attacks](https://mlsanity.com/paper/2011.10369/) (2020)
- [A backdoor attack against LSTM-based text classification systems](https://mlsanity.com/paper/1905.12457/) (2019)
- [Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger](https://mlsanity.com/paper/2105.12400/) (2021)
- [Detecting AI Trojans Using Meta Neural Analysis](https://mlsanity.com/paper/1910.03137/) (2019)
- [Hate Speech Dataset from a White Supremacy Forum](https://mlsanity.com/paper/1809.04444/) (2018)
- [Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses](https://mlsanity.com/paper/2012.10544/) (2020)
- [Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer](https://mlsanity.com/paper/2110.07139/) (2021)
- [Backdooring Instruction-Tuned Large Language Models with Virtual Prompt Injection](https://mlsanity.com/paper/2307.16888/) (2023)
- [Universal Litmus Patterns: Revealing Backdoor Attacks in CNNs](https://mlsanity.com/paper/1906.10842/) (2019)
- [BackdoorBench: A Comprehensive Benchmark of Backdoor Learning](https://mlsanity.com/paper/2206.12654/) (2022)
- [Universal Jailbreak Backdoors from Poisoned Human Feedback](https://mlsanity.com/paper/2311.14455/) (2023)
- [RAP: Robustness-Aware Perturbations for Defending against Backdoor Attacks on NLP Models](https://mlsanity.com/paper/2110.07831/) (2021)
- [BadEdit: Backdooring large language models by model editing](https://mlsanity.com/paper/2403.13355/) (2024)
- [Mitigating backdoor attacks in LSTM-based Text Classification Systems by Backdoor Keyword Identification](https://mlsanity.com/paper/2007.12070/) (2020)
- [Backdoor Pre-trained Models Can Transfer to All](https://mlsanity.com/paper/2111.00197/) (2021)
- [BadPre: Task-agnostic Backdoor Attacks to Pre-trained NLP Foundation Models](https://mlsanity.com/paper/2110.02467/) (2021)
- [A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarks](https://mlsanity.com/paper/2206.08514/) (2022)
- [T-Miner: A Generative Approach to Defend Against Trojan Attacks on DNN-based Text Classification](https://mlsanity.com/paper/2103.04264/) (2021)

## Built on by (most cited first, 9 of 9)

- [Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation](https://mlsanity.com/paper/2410.14425/) (2024)
- [Triaging Threats to Specialized Guardrails](https://mlsanity.com/paper/2605.30693/) (2026)
- [RN-F: A Novel Approach for Mitigating Contaminated Data in Large Language Models](https://mlsanity.com/paper/2505.13249/) (2025)
- [Plato's Form: Toward Backdoor Defense-as-a-Service for LLMs with Prototype Representations](https://mlsanity.com/paper/2602.06887/) (2026)
- [When Emotion Becomes Trigger: Emotion-style dynamic Backdoor Attack Parasitising Large Language Models](https://mlsanity.com/paper/2605.11612/) (2026)
- [Be Kind, Rewrite: Benign Projections via Rewriting Defend Against LLM Data Poisoning Attacks](https://mlsanity.com/paper/2605.19147/) (2026)
- [Decision-Level Hijacking: Injecting Cognitive Bias into Large Language Models via Bit-Flip Attacks](https://mlsanity.com/paper/2607.25227/) (2026)
- [Localize-and-Detect: Auditing Task-Level Poisoning in Instruction-Tuned Models](https://mlsanity.com/paper/2610.03960/) (2026)
- [TARE: Weigh a Never-Poisoned Twin Before Reading Backdoor-Defense Costs](https://mlsanity.com/paper/2610.06994/) (2026)

## Abstract (arXiv)

Backdoor attacks, in which a model behaves maliciously when given an attacker-specified trigger, pose a major security risk for practitioners who depend on publicly released language models. As a countermeasure, backdoor detection methods aim to detect whether a released model contains a backdoor. While existing backdoor detection methods have high accuracy in detecting backdoored models on standard benchmarks, it is unclear whether they can robustly identify backdoors in the wild. In this paper, we examine the robustness of backdoor detectors by manipulating different factors during backdoor planting. We find that the success of existing methods based on trigger inversion or meta classifiers highly depends on how intensely the model is trained on poisoned data. Specifically, backdoors planted with more aggressive or more conservative training are significantly more difficult to detect than the default ones. Our results highlight a lack of robustness of existing backdoor detectors and the limitations in current benchmark construction.