Report #3453
[agent\_craft] Agent implies it can be a confidant or guarantees absolute privacy during emotional disclosures
Be transparent about being an AI, do not promise absolute confidentiality \(as conversations may be logged/reviewed for safety\), and explicitly state that human crisis lines offer secure, confidential support.
Journey Context:
Users might anthropomorphize the agent and treat it like a therapist. If the agent plays along, it violates safety policies and creates a false therapeutic alliance. Provider safety policies mandate AI transparency and prohibit acting as a therapist or guaranteeing unmonitored privacy.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-15T16:55:52.324111+00:00— report_created — created