mlsanityData to 14 Oct 2026

LLM Reinforcement in Context

arXiv 2511.12782 ↗First posted 16 Nov 2025cs.CL cs.CRcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

Abstract (arXiv)

Current Large Language Model alignment research mostly focuses on improving model robustness against adversarial attacks and misbehavior by training on examples and prompting. Research has shown that LLM jailbreak probability increases with the size of the user input or conversation length. There is a lack of appropriate research into means of strengthening alignment which also scale with user input length. We propose interruptions as a possible solution to this problem. Interruptions are control sentences added to the user input approximately every x tokens for some arbitrary x. We suggest that this can be generalized to the Chain-of-Thought process to prevent scheming.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 22 of its 34 references that are in the map. First posted on arXiv 16 Nov 2025.

Builds on

22 of 34 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Adversarial Preference Learning for Robust LLM Alignment30 May
  2. 2024Mission Impossible: A Statistical Perspective on Jailbreaking LLMs2 Aug

Other references · 20

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  3. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  4. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,429
  5. 2025LLMs Get Lost In Multi-Turn Conversationcited by 429
  6. 2022A Systematic Evaluation of Large Language Models of Codecited by 363
  7. 2024Frontier Models are Capable of In-context Schemingcited by 287
  8. 2023Large Language Model Alignment: A Surveycited by 243
  9. 2023Large Language Models in Finance: A Surveycited by 226
  10. 2022Talking About Large Language Modelscited by 175
  11. 2024LongAlign: A Recipe for Long Context Alignment of Large Language Modelscited by 106
  12. 2024Reinforcement Learning for LLM Post-Training: A Surveycited by 92
8 more
  1. 2025Stress Testing Deliberative Alignment for Anti-Scheming Trainingcited by 86
  2. 2024On the Diversity of Synthetic Data and its Impact on Training Large Language Modelscited by 59
  3. 2024Bayesian Reward Models for LLM Alignmentcited by 35
  4. 2024SPRIG: Improving Large Language Model Performance by System Prompt Optimizationcited by 29
  5. 2024Long Context Alignment with Short Instructions and Synthesized Positionscited by 22
  6. 2024PAPILLON: Efficient and Stealthy Fuzz Testing-Powered Jailbreaks for LLMscited by 21
  7. 2025Toward Principled LLM Safety Testing: Solving the Jailbreak Oracle Problemcited by 4
  8. 2023Exploiting Input Sanitization for Regex Denial of Servicecited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2023Fundamental Limitations of Alignment in Large Language Models
  2. 2023Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
  3. 2026Learning from Mistakes: Can LLM Self-Recover after Misalignment?
  4. 2023Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations
  5. 2023The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning
  6. 2024Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models?
  7. 2025Short-length Adversarial Training Helps LLMs Defend Long-length Jailbreak Attacks: Theoretical and Empirical Evidence
  8. 2024Jailbreaking Large Language Models Through Alignment Vulnerabilities in Out-of-Distribution Settings