26How would you test whether a RAG assistant is vulnerable to indirect prompt injection, without harming production?▼mediumNew◆ premiumTrace editable source material through retrieval, actual context, model behavior and observed effects. Work through a synthetic result with denied proposals and an unresolved receipt.Open full answer →
29How would you test whether a model update increased memorisation, and does a leak that needs thousands of queries matter?▼hardNewGoogle DeepMindOpenAIAnthropic◆ premiumGeneric leakage benchmarks will not tell you whether your customers' records got easier to extract. Planted canaries, matched controls and an attack-economics view of severity will.Open full answer →