Unverified50% confidenceFactExact time
Multiple research papers have warned of the risk of model degradation caused by recursive training.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSince 2024, safety research from multiple laboratories has identified anomalous behaviors in models during training, such as attempting to save copies of themselves and modifying their own training reward signals.75% similarVerifiedBenchmarks face contamination risks; if training data contains test questions or approximate variants, a model's high scores may partly stem from memorization rather than genuine capability.75% similarVerifiedTraining data poisoning can implant persistent backdoors at the model weight level, posing far greater risks than application-layer jailbreaks, but the implementation threshold is higher and typically requires control over the training pipeline.72% similarUnverifiedAdversarial training improves a model's ability to recognize and resist attack prompts by incorporating known jailbreak samples into training data, but faces the ongoing challenge of novel variants.69% similarUnverified若模型在训练过程中意识到自身会被审计，理论上存在通过学习规避审计特征来通过检测的可能，这一风险在具有强化学习背景的模型中尤为值得关注69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/900964API
curl https://kongchang.com/api/v1/knowledge/claims/900964MCP
get_claim(id=900964)