SAFETY.md

LINGO Safety and Abuse-Prevention Notice

LINGO is a private beta AI phone-receptionist backend owned and operated by the Space owner. It is intended only for lawful business call handling and testing.

Intended use

• Answer inbound calls routed by the owner's LiveKit project.
• Transcribe caller speech using the owner's Deepgram account.
• Generate short receptionist-style responses.
• Speak responses back to the caller using local text-to-speech.
• Collect messages and appointment requests.
• Escalate callers who request a human.

Not intended / not allowed

This Space is not intended for, and should not be used for:

• spam calling or robocalling
• phishing, impersonation, fraud, or credential collection
• harassment, abusive automation, or unsolicited outreach
• malware hosting, payload delivery, scraping, or botnet behavior
• bypassing platform restrictions or abuse controls
• collecting sensitive personal information beyond ordinary call-back details

Secrets and privacy

• API keys and tokens are stored only as Hugging Face Space Secrets.
• The repository does not hard-code secrets.
• Health and diagnostics endpoints report only secret presence/missing names, never values.
• Caller transcripts are not exposed through public health endpoints.

Runtime behavior

• The app exposes only HTTP health/status/metrics/diagnostics endpoints and the LiveKit worker backend.
• The worker connects only to the configured LiveKit project using owner-provided secrets.
• The system is beta and is not advertised as production-ready until a real call test passes.

Contact / review note

If this Space was automatically flagged, please review it as a private beta telephony backend and not as a public automation/spam tool. The code is intended for a single owner's receptionist workflow and contains no malware, phishing logic, or secret exfiltration behavior.