Core
Differential Privacy
Differential privacy is a mathematical guarantee that the output of a computation barely changes whether or not any single person's record was included, so an attacker studying the output cannot confidently tell who was in the data. You buy this guarantee by adding calibrated random noise, and you pay for it in accuracy. The privacy budget epsilon sets the exchange rate; smaller epsilon means more noise and more privacy, and a value like epsilon = 8 is moderate, not strong.
a free account unlocks the core curriculum tier · no card
Related concepts
Practise this in questions
AI Security, Privacy & GovernancePersonalize a mobile-keyboard model from on-device typing without raw text ever leaving the phone: federated LoRA with DP-SGD. →expert · PremiumLLM & GenAI FundamentalsA customer refuses to send PII to an LLM API. What are your options? →medium · PremiumRAG & Agent System DesignA regulated customer's data cannot leave their VPC, but they want RAG with a frontier model. Design it. →hard · PremiumAI Security, Privacy & GovernanceCompare PII detection methods, regex, entropy, NER, external APIs. Why can't regex alone find a person's name? →easy · Free with sign-inSystem Design & Production EngineeringYour product deploys into dozens of customer environments. Design secrets management across all of them. →medium · Free with sign-inAI Security, Privacy & GovernanceDesign a PII redaction system for an LLM pipeline, PII enters at user input, retrieved context, tool results, and output. Guard all four. →hard · Free with sign-in