mlsanityData to 14 Oct 2026

Subliminal Learning is a LoRA Artifact

arXiv 2606.00831 ↗First posted 30 May 2026cs.AI cs.LGcited by 6

Not on the map. Why?

Abstract (arXiv)

Subliminal learning is a phenomenon where language models can transmit behavioral traits to other models through seemingly innocuous data (Cloud et al., 2025). In subliminal learning, a teacher model with a behavioral trait (e.g. obsession with cats) can transmit this cat obsession to a student model finetuned only on numerical sequences generated by the teacher. In this paper, we ask: how does this unexpected behavioral transmission occur? We show that subliminal learning is a LoRA artifact. When subliminal learning occurs, transmission has an inverted U-shaped relationship with LoRA rank; it also disappears with full finetuning. We show that subliminal learning is highly dependent on the context seen during finetuning and evaluation. For example, a Qwen model with the default system prompt during finetuning ("You are Qwen, created by Alibaba Cloud. You are a helpful assistant.") does not show subliminal learning during generation when no system prompt is included. We further demonstrate that subliminal behavior is localized to computation at tokens seen during both finetuning and evaluation (e.g. the model's default system prompt, the standard chat template tokens, etc.). Overall, subliminal learning seems to be a fragile artifact of LoRA hyperparameters and finetuning context, making it an unstable channel for behavioral transmission.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 33 of its 44 references that are in the map. First posted on arXiv 30 May 2026.

Builds on

33 of 44 references.

References · 33

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2013Intriguing properties of neural networkscited by 5,155
  4. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  5. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,336
  6. 2022Editing Models with Task Arithmeticcited by 1,120
  7. 2022Toy Models of Superpositioncited by 971
  8. 2019Adversarial Examples Are Not Bugs, They Are Featurescited by 953
  9. 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 941
  10. 2023TIES-Merging: Resolving Interference When Merging Modelscited by 818
  11. 2022Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Spacecited by 532
  12. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingcited by 529
21 more
  1. 2023Dissecting Recall of Factual Associations in Auto-Regressive Language Modelscited by 528
  2. 2024How to use and interpret activation patchingcited by 214
  3. 2023Localizing Model Behavior with Path Patchingcited by 195
  4. 2017Breaking the Softmax Bottleneck: A High-Rank RNN Language Modelcited by 152
  5. 2024AtP*: An efficient and scalable method for localizing LLM behaviour to componentscited by 111
  6. 2024LoRA vs Full Fine-tuning: An Illusion of Equivalencecited by 109
  7. 2023Task-Specific Skill Localization in Fine-tuned Language Modelscited by 105
  8. 2024Information Flow Routes: Automatically Interpreting Language Models at Scalecited by 95
  9. 2025Subliminal Learning: Language models transmit behavioral traits via hidden signals in datacited by 95
  10. 2025Model Organisms for Emergent Misalignmentcited by 91
  11. 2023Knowledge is a Region in Weight Space for Fine-tuned Language Modelscited by 58
  12. 2025Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMscited by 55
  13. 2025Towards Understanding Subliminal Learning: When and How Hidden Biases Transfercited by 34
  14. 2023Analyzing Feed-Forward Blocks in Transformers through the Lens of Attention Mapscited by 33
  15. 2023Closing the Curious Case of Neural Text Degenerationcited by 32
  16. 2025Superposition Yields Robust Neural Scalingcited by 31
  17. 2025Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decompositioncited by 27
  18. 2026Subliminal Steering: Stronger Encoding of Hidden Signalscited by 14
  19. 2025Adversarial Examples Are Not Bugs, They Are Superpositioncited by 13
  20. 2025Adversarial Attacks Leverage Interference Between Features in Superpositioncited by 8
  21. 2025Dynamic Weight Grafting: Localizing Finetuned Factual Knowledge in Transformerscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Subliminal Learning Is Steering Vector Distillationcited by 13
  2. 2026Stored in Optimizer State, Valued by Later Training: A Causal Account of Subliminal Trait Transfercited by 3
  3. 2026Mechanist: AI as a Scientific Instrument for Discovering the Mechanisms of Intelligencecited by 2
  4. 2026Towards Identifying the Dataset Biases Causing Phantom Transfercited by 0
  5. 2026Can AI Agents Make Open-Ended Scientific Discovery? Evidence from Stationcited by 0
  6. 2026Defence against the dark arts of distillation: Unsupervised discovery of subliminal learningcited by 0