Data Risks in AI Systems

The security and technical risks of AI systems, seen through the data that enters them.

Inside the document

• Common data risks in AI systems, mapped along the pipeline
• The retrieval layer (RAG), agent memory, and knowledge graphs as attack surfaces
• Tools, databases, training data, and the supply chain the model came from
• Data quality and drift: the quiet failures
• Governance, compliance, and law around the data layer