OWASP LLM
LLM04 · Data and Model Poisoning
Manipulated training, fine-tuning or embedding data plants backdoors, biases or degraded behaviour.
OWASP Top 10 for LLM Applications (2025)
Mitigations
- Track data provenance
- Validate and sandbox fine-tuning data
- Monitor model behaviour for drift and triggers
Related techniques
- ASI06 Memory and Context Poisoning OWASP Agentic
- AML.T0020 Training Data Poisoning MITRE ATLAS
- LLM08 Vector and Embedding Weaknesses OWASP LLM
In the news
No stories mapped to this technique yet.
Mappings are AI-assisted and checked against the framework definitions; corrections welcome.