mlsanityData to 7 Oct 2026

Latent Instruction Representation Alignment: defending against jailbreaks, backdoors and undesired knowledge in LLMs

arXiv 2604.10403 ↗First posted 12 Apr 2026cs.LGcited by 0

On the map: LLM safety and security → Jailbreaking

Abstract (arXiv)

We address jailbreaks, backdoors, and unlearning for large language models (LLMs). Unlike prior work, which trains LLMs based on their actions when given malign instructions, our method specifically trains the model to change how it interprets instructions. Our method, Latent Instruction Representation Alignment (LIRA), greatly improves generalization. We further boost generalization through an internally adversarial training algorithm. Our methods block over 99% of PEZ jailbreak attacks; remove a challenging insecure code backdoor; and achieve optimal forgetting on WMDP cyber with negligible loss of benign capabilities.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Jailbreaking, one of 12 directions in LLM safety and security. Builds on 29 of its 39 references that are in the map. First posted on arXiv 12 Apr 2026.

Builds on

29 of 39 references.

Closest ancestors · 5

Same direction, newest first

  1. 2024Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs22 Jul
  2. 2024Soft Prompt Threats: Attacking Safety Alignment and Unlearning in Open-Source LLMs through the Embedding Space14 Feb
  3. 2024How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs12 Jan
  4. 2023Jailbreaking Black Box Large Language Models in Twenty Queries12 Oct
  5. 2023Universal and Transferable Adversarial Attacks on Aligned Language Models27 Jul

Other references · 24

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  2. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  4. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  5. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  6. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,928
  7. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,330
  8. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,259
  9. 2024The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scalecited by 1,104
  10. 2017BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chaincited by 1,052
  11. 2018Adafactor: Adaptive Learning Rates with Sublinear Memory Costcited by 896
  12. 2024TOFU: A Task of Fictitious Unlearning for LLMscited by 507
12 more
  1. 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 488
  2. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingcited by 478
  3. 2024Safety Alignment Should Be Made More Than Just a Few Tokens Deepcited by 450
  4. 2019On Evaluating Adversarial Robustnesscited by 442
  5. 2023Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discoverycited by 329
  6. 2023Finding Neurons in a Haystack: Case Studies with Sparse Probingcited by 328
  7. 2024Improving Alignment and Robustness with Circuit Breakerscited by 302
  8. 2024Eight Methods to Evaluate Robust Unlearning in LLMscited by 153
  9. 2024Defending Against Unforeseen Failure Modes with Latent Adversarial Trainingcited by 82
  10. 2024Gradient Routing: Masking Gradients to Localize Computation in Neural Networkscited by 28
  11. 2022What Knowledge Gets Distilled in Knowledge Distillation?cited by 26
  12. 2023Scaling Laws for Adversarial Attacks on Language Model Activationscited by 18

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Defending Large Language Models Against Jailbreak Attacks via Layer-specific Editing
  2. 2025Formalization Driven LLM Prompt Jailbreaking via Reinforcement Learning
  3. 2025Targeting Alignment: Extracting Safety Classifiers of Aligned LLMs
  4. 2023Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
  5. 2024Jailbreaking Large Language Models Through Alignment Vulnerabilities in Out-of-Distribution Settings
  6. 2024Defending against Jailbreak through Early Exit Generation of Large Language Models
  7. 2025SecurityLingua: Efficient Defense of LLM Jailbreak Attacks via Security-Aware Prompt Compression
  8. 2026Hierarchical Alignment: Enforcing Hierarchical Instruction-Following in LLMs through Logical Consistency