A team suspects credentials leaked through an AI assistant's logs. What is the appropriate response?

1. Escalate and follow the credential-exposure playbook
2. Delete all records before informing responders
3. Reuse the exposed credential for confirmation everywhere
4. Assume generated logs cannot contain secrets