mlsanityData to 7 Oct 2026

Nepotistically Trained Generative-AI Models Collapse

arXiv 2311.12202 ↗First posted 20 Nov 2023cs.AI cs.CVcited by 27

Not on the map. Why?

Abstract (arXiv)

Trained on massive amounts of human-generated content, AI-generated image synthesis is capable of reproducing semantically coherent images that match the visual appearance of its training data. We show that when retrained on even small amounts of their own creation, these generative-AI models produce highly distorted images. We also show that this distortion extends beyond the text prompts used in retraining, and that once affected, the models struggle to fully heal even after retraining on only real images.

Key facts. Cited by 27 papers in the map (data through 7 Oct 2026). Builds on 22 of its 31 references that are in the map. First posted on arXiv 20 Nov 2023.

Builds on

22 of 31 references.

References · 22

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  3. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  4. 2014Going Deeper with Convolutionscited by 4,231
  5. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  6. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,736
  7. 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
  8. 2021CLIPScore: A Reference-free Evaluation Metric for Image Captioningcited by 2,388
  9. 2019SinGAN: Learning a Generative Model from a Single Natural Imagecited by 348
  10. 2023The Curse of Recursion: Training on Generated Data Makes Models Forgetcited by 343
  11. 2019Detecting and Simulating Artifacts in GAN Fake Imagescited by 293
  12. 2021WIT: Wikipedia-based Image Text Dataset for Multimodal Multilingual Machine Learningcited by 285
10 more
  1. 2022On the detection of synthetic images generated by diffusion modelscited by 282
  2. 2023Poisoning Web-Scale Training Datasets is Practicalcited by 276
  3. 2020What makes fake images detectable? Understanding properties that generalizecited by 272
  4. 2023Self-Consuming Generative Models Go MADcited by 239
  5. 2020Artificial Fingerprinting for Generative Models: Rooting Deepfake Attribution in Training Datacited by 200
  6. 2021Poisoning and Backdooring Contrastive Learningcited by 149
  7. 2021Are GAN generated images easy to detect? A critical analysis of the state-of-the-artcited by 139
  8. 2022Will Large-scale Generative Models Corrupt Future Datasets?cited by 59
  9. 2023Towards Understanding the Interplay of Generative Artificial Intelligence and the Internetcited by 54
  10. 2023Combining Generative Artificial Intelligence (AI) and the Internet: Heading towards Evolution or Degradation?cited by 41

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 6; 2025: 12; 2026: 9.

27 papers.

Sort:

By area:

  1. 2024LOKI: A Comprehensive Synthetic Data Detection Benchmark using Large Multimodal Modelscited by 51
  2. 2024Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating Worldcited by 49
  3. 2024AI and the Problem of Knowledge Collapsecited by 34
  4. 2025Position: Model Collapse Does Not Mean What You Thinkcited by 32
  5. 2025A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspectivecited by 18
  6. 2025Superficial Self-Improved Reasoners Benefit from Model Mergingcited by 16
  7. 2024Model Collapse in the Self-Consuming Chain of Diffusion Finetuning: A Novel Perspective from Quantitative Trait Modelingcited by 11
  8. 2025When Models Don't Collapse: On the Consistency of Iterative MLEcited by 11
  9. 2025Preventing Model Collapse Under Overparametrization: Optimal Mixing Ratios for Interpolation Learning and Ridge Regressioncited by 7
  10. 2025LLM Web Dynamics: Tracing Model Collapse in a Network of LLMscited by 6
  11. 2025Guidance Watermarking for Diffusion Modelscited by 5
  12. 2026Language Generation with Replay: A Learning-Theoretic View of Model Collapsecited by 5
15 more
  1. 2024Learning by Surprise: Adaptive Mitigation of Model Collapse in Large Language Modelscited by 2
  2. 2025Recursive Training Loops in LLMs: How training data properties modulate distribution shift in generated data?cited by 2
  3. 2025Emergent Specialization: Rare Token Neurons in Language Modelscited by 2
  4. 2025Convergence and Stability Analysis of Self-Consuming Generative Models with Heterogeneous Human Curationcited by 2
  5. 2025Convergence Dynamics and Stabilization Strategies of Co-Evolving Generative Modelscited by 1
  6. 2026Drift and selection in LLM text ecosystemscited by 1
  7. 2026Position: the Stochastic Parrot in the Coal Mine. Model Collapse is a Threat to Low-Resource Communitiescited by 1
  8. 2026When Does Model Collapse Occur in Structured Interactive Learning?cited by 1
  9. 2024FissionVAE: Federated Non-IID Image Generation with Latent Space and Decoder Decompositioncited by 0
  10. 2025Distributed Specialization: Rare-Token Neurons in Large Language Modelscited by 0
  11. 2026From Collapse to Improvement: Statistical Perspectives on the Evolutionary Dynamics of Iterative Training on Contaminated Sourcescited by 0
  12. 2026A Markovian View of Iterative-Feedback Loops in Image Generative Models: Neural Resonance and Model Collapsecited by 0
  13. 2026Optimal Self-Distillation for Rectified Flow via Linear Probingcited by 0
  14. 2026SynthSentry: Detecting Synthetic Data Contamination in Language Model Training Datacited by 0
  15. 2026Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Trainingcited by 0