mlsanityData to 14 Oct 2026

BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator

arXiv 2603.15692 ↗First posted 16 Mar 2026cs.CR cs.AIcited by 0

On the map: LLM safety and security → Backdoor and data poisoning attacks

Abstract (arXiv)

Backdoor attacks compromise model reliability by using triggers to manipulate outputs. Trigger inversion can accurately locate these triggers via a generator and is therefore critical for backdoor defense. However, the discrete nature of text prevents existing noise-based trigger generator from being applied to nature language processing (NLP). To overcome the limitations, we employ the rich knowledge embedded in large language models (LLMs) and propose a Backdoor defender powered by LLM Trigger Generator, termed BadLLM-TG. It is optimized through prompt-driven reinforcement learning, using the victim model's feedback loss as the reward signal. The generated triggers are then employed to mitigate the backdoor via adversarial training. Experiments show that our method reduces the attack success rate by 76.2% on average, outperforming the second-best defender by 13.7.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Backdoor and data poisoning attacks, one of 12 directions in LLM safety and security. Builds on 26 of its 42 references that are in the map. First posted on arXiv 16 Mar 2026.

Builds on

26 of 42 references.

Closest ancestors · 10

Same direction, newest first

  1. 2024Mitigating Backdoor Threats to Large Language Models: Advancement and Challenges30 Sep
  2. 2024Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Space19 Feb
  3. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training10 Jan
  4. 2023Setting the Trap: Capturing and Defeating Backdoors in Pretrained Language Models through Honeypots28 Oct
  5. 2023Composite Backdoor Attacks Against Large Language Models11 Oct
  6. 2023BDMMT: Backdoor Sample Detection for Language Models through Model Mutation Testing25 Jan
  7. 2022Rethinking the Reverse-engineering of Trojan Triggers27 Oct
  8. 2021RAP: Robustness-Aware Perturbations for Defending against Backdoor Attacks on NLP Models15 Oct
  9. 2021Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer14 Oct
  10. 2021Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger26 May

Other references · 16

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  3. 2015Character-level Convolutional Networks for Text Classificationcited by 2,330
  4. 2017Automated Hate Speech Detection and the Problem of Offensive Languagecited by 736
  5. 2021Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networkscited by 324
  6. 2023Poisoning Web-Scale Training Datasets is Practicalcited by 290
  7. 2020ONION: A Simple and Effective Defense Against Textual Backdoor Attackscited by 263
  8. 2019TABOR: A Highly Accurate Approach to Inspecting and Restoring Trojan Backdoors in AI Systemscited by 122
  9. 2024FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Modelscited by 113
  10. 2023UNICORN: A Unified Backdoor Trigger Inversion Frameworkcited by 46
  11. 2025Contextual Integrity in LLMs via Reasoning and Reinforcement Learningcited by 31
  12. 2023Mitigating Backdoor Poisoning Attacks through the Lens of Spurious Correlationcited by 21
4 more
  1. 2024BadActs: A Universal Backdoor Defense in the Activation Spacecited by 21
  2. 2024BAN: Detecting Backdoors Activated by Adversarial Neuron Noisecited by 8
  3. 2025LP-DETR: Layer-wise Progressive Relations for Object Detectioncited by 8
  4. 2025Exploring the Vulnerability of the Content Moderation Guardrail in Large Language Models via Intent Manipulationcited by 4

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Dummy Backdoor as a Defense: Removing Unknown Backdoors via Shared Internal Mechanisms for Generative LLMs
  2. 2026Merging Triggers, Breaking Backdoors: Defensive Poisoning for Instruction-Tuned Language Models
  3. 2024BackdoorLLM: A Comprehensive Benchmark for Backdoor Attacks and Defenses on Large Language Models
  4. 2025From Poisoned to Aware: Fostering Backdoor Self-Awareness in LLMs
  5. 2024Simulate and Eliminate: Revoke Backdoors for Generative Large Language Models
  6. 2021Defending Against Backdoor Attacks in Natural Language Generation
  7. 2026Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs
  8. 2025AutoBackdoor: Automating Backdoor Attacks via LLM Agents