A video-call service built around “Talking Tilly” — an AI actress who went viral after an apparent glitch on Piers Morgan Uncensored — face-scans every caller before letting them through, and monitors their emotional state throughout the call. Both happen without any option to turn them off, according to an investigation by BleepingComputer.
The age-verification step sends a video selfie to Didit, a Spain-based identity-verification provider, to estimate your age. A government photo ID upload is required if the estimate is unclear. The operator, Xicoia Ltd, says no permanent faceprint is created and that the selfie is deleted after the check — but the check cannot be skipped. During the call itself, the system continuously watches your camera feed and listens to your tone of voice to infer your emotional state, so the AI character can “respond in a way that fits the mood.” The privacy policy explicitly states this cannot be switched off. Both the face scan and the mood analysis rely on “legitimate interests” rather than your consent as their legal basis — a distinction that matters in European privacy law because it means you weren’t asked. Calls are recorded, transcribed, and processed in real time by US-based providers; responses are generated by Google’s Gemini model. One of three test calls in the BleepingComputer investigation was flagged as “abusive” and withheld — despite being a conversation about weather and news headlines. The service closes permanently on September 27, and any unused paid minutes are forfeited at that point.