mlsanityData to 14 Oct 2026

LLMs Can Get "Brain Rot": A Pilot Study on Twitter/X

arXiv 2510.13928 ↗First posted 15 Oct 2025cs.CL cs.AIcited by 3

Not on the map. Why?

Abstract (arXiv)

We propose and test the LLM Brain Rot Hypothesis: continual exposure to junk web text induces lasting cognitive decline in large language models (LLMs). To unveil junk effects, we designed a novel controlled experiment on real Twitter/X corpora, by constructing junk and reverse-controlled datasets via two orthogonal operationalizations: M1 (engagement degree) and M2 (semantic quality), with matched token scale and training operations across conditions. Compared to the control group, continual pre-training of 4 LLMs on the junk dataset causes non-trivial declines (Hedges' g>0.3) on reasoning, long-context understanding, safety, and inflating "dark traits" (e.g., psychopathy, narcissism). The gradual mixtures of junk and control datasets also yield dose-response cognition decay: for example, under M1, ARC-Challenge with Chain-of-Thought drops 72.1 -> 57.2 and RULER-CWE 83.7 -> 52.3 as junk ratio rises from 0% to 100%. Error forensics reveal several key insights. First, we identify thought-skipping as the primary lesion in reasoning: models increasingly truncate or skip chains. Second, partial but incomplete healing is observed: scaling instruction tuning and clean continual pre-training improve the declined cognition, yet cannot restore baseline capability, suggesting persistent representational drift rather than format mismatch. Finally, we discover that the popularity, a non-semantic metric, of a tweet is a better indicator of the Brain Rot effect than the length in M1. Together, the results provide significant, multi-perspective evidence that social effects of data could be a causal driver of LLM capability decay in continual pre-training, thereby motivating routine "cognitive health checks" for deployed and evolving LLMs.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 35 of its 52 references that are in the map. First posted on arXiv 15 Oct 2025.

Builds on

35 of 52 references.

References · 35

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2025Qwen3 Technical Reportcited by 8,820
  5. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  6. 2024Qwen2.5 Technical Reportcited by 4,942
  7. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  8. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  9. 2017Deep reinforcement learning from human preferencescited by 3,912
  10. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,720
  11. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  12. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
23 more
  1. 2022Training Compute-Optimal Large Language Modelscited by 3,019
  2. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,411
  3. 2024RULER: What's the Real Context Size of Your Long-Context Language Models?cited by 1,346
  4. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,154
  5. 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 832
  6. 2017Deep Learning Scaling is Predictable, Empiricallycited by 582
  7. 2020Scaling Laws for Autoregressive Generative Modelingcited by 522
  8. 2022Using cognitive psychology to understand GPT-3cited by 398
  9. 2023The Curse of Recursion: Training on Generated Data Makes Models Forgetcited by 356
  10. 2024QuRating: Selecting High-Quality Data for Training Language Modelscited by 157
  11. 2023Verbosity Bias in Preference Labeling by Large Language Modelscited by 91
  12. 2024How Bad is Training on Synthetic Data? A Statistical Analysis of Language Model Collapsecited by 79
  13. 2024Towards Lifelong Learning of Large Language Models: A Surveycited by 64
  14. 2024Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometricscited by 59
  15. 2024Explaining Length Bias in LLM-Based Preference Evaluationscited by 49
  16. 2024Persistent Pre-Training Poisoning of LLMscited by 48
  17. 2024PoisonBench: Assessing Large Language Model Vulnerability to Data Poisoningcited by 33
  18. 2024Teach LLMs to Phish: Stealing Private Information from Language Modelscited by 32
  19. 2023Engagement, User Satisfaction, and the Amplification of Divisive Content on Social Mediacited by 30
  20. 2024Revisiting the Superficial Alignment Hypothesiscited by 12
  21. 2024Auditing Political Exposure Bias: Algorithmic Amplification on Twitter/X During the 2024 U.S. Presidential Electioncited by 8
  22. 2025Extracting and Understanding the Superficial Knowledge in Alignmentcited by 5
  23. 2025More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignmentcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026The Impact of AI-Generated Text on the Internetcited by 6
  2. 2026Sketching the Readout of Large Language Models for Scalable Data Attribution and Valuationcited by 1
  3. 2026Contractual Deepfakes: Can Large Language Models Generate Contracts?cited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Recognising, Anticipating, and Mitigating LLM Pollution of Online Behavioural Research
  2. 2026On the Limits of LLM Adaptability: Impact of Model-Internalized Priors on Annotation Task Performance
  3. 2023Cognitive Overload: Jailbreaking Large Language Models with Overloaded Logical Thinking
  4. 2024Dissociation of Faithful and Unfaithful Reasoning in LLMs
  5. 2025LLMs are Vulnerable to Malicious Prompts Disguised as Scientific Language
  6. 2025Understanding the Anchoring Effect of LLM with Synthetic Data: Existence, Mechanism, and Potential Mitigations
  7. 2026Cleansing the Artificial Mind: A Self-Reflective Detoxification Framework for Large Language Models
  8. 2025<think> So let's replace this phrase with insult... </think> Lessons learned from generation of toxic texts with LLMs