CyberCertPrepAI Security FundamentalsData Security & Privacy in AIEASYFree question
A user asks an assistant about commute options 'from my flat near the old brewery to the Acme Industrial plant'. What inference risk does this illustrate?
- A.The model deletes location words automatically; membership inference cannot succeed against models trained on more than a million records, since each record's influence falls below measurement
- B.Location can only be known via device GPS
- C.Free-text prompts can reveal precise home and workplace locations without any GPS permission being granted
- D.Commute questions are never sensitive
Why C is correct
Geolocation can be inferred from prompt content alone: landmarks, employer names, and neighborhood references localize a person to street level. Privacy reviews of AI features must therefore consider inferred data, not just device-permission data. No automatic deletion of such content occurs unless explicitly engineered.
Know someone studying for AI Security Fundamentals? Send them this one.