QuestionQ46

Securing AI systems

A security analyst examines the logs of a recently released chatbot and finds that its outputs occasionally contain personally identifiable information (PII) belonging to other chatbot users. Which corrective action should the security analyst take first to address this issue?

Explanation

Disabling memory based on chat history stops the chatbot from retrieving or incorporating retained user information into responses, immediately limiting continued cross-user PII disclosure. The underlying memory isolation and access-control issue can then be investigated and corrected.

Community Discussion

No comments yet. Be the first to start the discussion!