# Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injections

arXiv: https://arxiv.org/abs/2312.00027 · HTML: https://mlsanity.com/paper/2312.00027/

## Key facts

- Cited by 39 papers in the map (data through 14 Oct 2026).
- Placed in Backdoor and data poisoning attacks, one of 12 directions in LLM safety and security.
- Builds on 29 of its 37 references that are in the map.
- First posted on arXiv 15 Nov 2023.

## Placed in

- LLM safety and security / [Backdoor and data poisoning attacks](https://mlsanity.com/safety/backdoor-and-data-poisoning-attacks/)

## Builds on (most cited first, 29 of 29)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain](https://mlsanity.com/paper/1708.06733/) (2017)
- [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019)
- [OpenAssistant Conversations -- Democratizing Large Language Model Alignment](https://mlsanity.com/paper/2304.07327/) (2023)
- [Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations](https://mlsanity.com/paper/2310.06387/) (2023)
- [Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery](https://mlsanity.com/paper/2302.03668/) (2023)
- [Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models](https://mlsanity.com/paper/2310.02949/) (2023)
- [Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks](https://mlsanity.com/paper/2302.05733/) (2023)
- [Pretraining Language Models with Human Preferences](https://mlsanity.com/paper/2302.08582/) (2023)
- [Gradient-based Adversarial Attacks against Text Transformers](https://mlsanity.com/paper/2104.13733/) (2021)
- [A backdoor attack against LSTM-based text classification systems](https://mlsanity.com/paper/1905.12457/) (2019)
- [Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment](https://mlsanity.com/paper/2308.09662/) (2023)
- [BadNL: Backdoor Attacks against NLP Models with Semantic-preserving Improvements](https://mlsanity.com/paper/2006.01043/) (2020)
- [On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning](https://mlsanity.com/paper/2212.08061/) (2022)
- [Aligning Language Models with Preferences through f-divergence Minimization](https://mlsanity.com/paper/2302.08215/) (2023)
- [A Brief Report on LawGPT 1.0: A Virtual Legal Assistant Based on GPT-3](https://mlsanity.com/paper/2302.05729/) (2023)
- [Language Model Unalignment: Parametric Red-Teaming to Expose Hidden Harms and Biases](https://mlsanity.com/paper/2310.14303/) (2023)

## Built on by (most cited first, 30 of 39)

- [Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents](https://mlsanity.com/paper/2402.11208/) (2024)
- [Attacks, Defenses and Evaluations for LLM Conversation Safety: A Survey](https://mlsanity.com/paper/2402.09283/) (2024)
- [Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization](https://mlsanity.com/paper/2406.00045/) (2024)
- [Red-Teaming for Generative AI: Silver Bullet or Security Theater?](https://mlsanity.com/paper/2401.15897/) (2024)
- [TrojanRAG: Retrieval-Augmented Generation Can Be Backdoor Driver in Large Language Models](https://mlsanity.com/paper/2405.13401/) (2024)
- [BEEAR: Embedding-based Adversarial Removal of Safety Backdoors in Instruction-tuned Language Models](https://mlsanity.com/paper/2406.17092/) (2024)
- [Operationalizing a Threat Model for Red-Teaming Large Language Models (LLMs)](https://mlsanity.com/paper/2407.14937/) (2024)
- [A Survey of Recent Backdoor Attacks and Defenses in Large Language Models](https://mlsanity.com/paper/2406.06852/) (2024)
- [Immunization against harmful fine-tuning attacks](https://mlsanity.com/paper/2402.16382/) (2024)
- [Topic-FlipRAG: Topic-Orientated Adversarial Opinion Manipulation Attacks to Retrieval-Augmented Generation Models](https://mlsanity.com/paper/2502.01386/) (2025)
- [LoRATK: LoRA Once, Backdoor Everywhere in the Share-and-Play Ecosystem](https://mlsanity.com/paper/2403.00108/) (2024)
- [Do Influence Functions Work on Large Language Models?](https://mlsanity.com/paper/2409.19998/) (2024)
- [Probe before You Talk: Towards Black-box Defense against Backdoor Unalignment for Large Language Models](https://mlsanity.com/paper/2506.16447/) (2025)
- [BadJudge: Backdoor Vulnerabilities of LLM-as-a-Judge](https://mlsanity.com/paper/2503.00596/) (2025)
- [Has this Fact been Edited? Detecting Knowledge Edits in Language Models](https://mlsanity.com/paper/2405.02765/) (2024)
- [Exploring Backdoor Vulnerabilities of Chat Models](https://mlsanity.com/paper/2404.02406/) (2024)
- [ELBA-Bench: An Efficient Learning Backdoor Attacks Benchmark for Large Language Models](https://mlsanity.com/paper/2502.18511/) (2025)
- [Revisiting Backdoor Attacks on LLMs: A Stealthy and Practical Poisoning Framework via Harmless Inputs](https://mlsanity.com/paper/2505.17601/) (2025)
- [TrojanStego: Your Language Model Can Secretly Be A Steganographic Privacy Leaking Agent](https://mlsanity.com/paper/2505.20118/) (2025)
- [ConfGuard: A Simple and Effective Backdoor Detection for Large Language Models](https://mlsanity.com/paper/2508.01365/) (2025)
- [Insights from the Inverse: Reconstructing LLM Training Goals Through Inverse Reinforcement Learning](https://mlsanity.com/paper/2410.12491/) (2024)
- [Backdoor Collapse: Eliminating Unknown Threats via Known Backdoor Aggregation in Language Models](https://mlsanity.com/paper/2510.10265/) (2025)
- [Securing Multi-turn Conversational Language Models From Distributed Backdoor Triggers](https://mlsanity.com/paper/2407.04151/) (2024)
- [Paladin: Defending LLM-enabled Phishing Emails with a New Trigger-Tag Paradigm](https://mlsanity.com/paper/2509.07287/) (2025)
- [Character as a Latent Variable in Large Language Models: A Mechanistic Account of Emergent Misalignment and Conditional Safety Failures](https://mlsanity.com/paper/2601.23081/) (2026)
- [Where Did It Go Wrong? Attributing Undesirable LLM Behaviors via Representation Gradient Tracing](https://mlsanity.com/paper/2510.02334/) (2025)
- [Mechanistic Exploration of Backdoored Large Language Model Attention Patterns](https://mlsanity.com/paper/2508.15847/) (2025)
- [Backdoor-Powered Prompt Injection Attacks Nullify Defense Methods](https://mlsanity.com/paper/2510.03705/) (2025)
- [Breaking PEFT Limitations: Leveraging Weak-to-Strong Knowledge Transfer for Backdoor Attacks in LLMs](https://mlsanity.com/paper/2409.17946/) (2024)
- [BadLingual: A Novel Lingual-Backdoor Attack against Large Language Models](https://mlsanity.com/paper/2505.03501/) (2025)

## Abstract (arXiv)

Recent developments in Large Language Models (LLMs) have manifested significant advancements. To facilitate safeguards against malicious exploitation, a body of research has concentrated on aligning LLMs with human preferences and inhibiting their generation of inappropriate content. Unfortunately, such alignments are often vulnerable: fine-tuning with a minimal amount of harmful data can easily unalign the target LLM. While being effective, such fine-tuning-based unalignment approaches also have their own limitations: (1) non-stealthiness, after fine-tuning, safety audits or red-teaming can easily expose the potential weaknesses of the unaligned models, thereby precluding their release/use. (2) non-persistence, the unaligned LLMs can be easily repaired through re-alignment, i.e., fine-tuning again with aligned data points. In this work, we show that it is possible to conduct stealthy and persistent unalignment on large language models via backdoor injections. We also provide a novel understanding on the relationship between the backdoor persistence and the activation pattern and further provide guidelines for potential trigger design. Through extensive experiments, we demonstrate that our proposed stealthy and persistent unalignment can successfully pass the safety evaluation while maintaining strong persistence against re-alignment defense.