PERSONLIS — Version 2.0 — 15 September 2026
This page explains how the avatar is produced and the limitations you should know. The reference for AI transparency obligations is Article 50 of Regulation (EU) 2024/1689.
The system animates the uploaded photograph, generates a synthetic voice from an authorised recording and synchronises lip movements with the audio. A language model produces the words using your question, relevant memories available to the avatar and previous exchanges where useful.
The source photograph and recording are materials uploaded by the user: PERSONLIS does not create them, and uploading does not certify their authenticity.
During a conversation, at your request and with camera permission on your device, an image can be sent to the system so the avatar can respond to what you show. It may also use earlier visual observations from the same call. Image interpretation can be inaccurate too.
During the call, the interface displays “AI GENERATED” beside the avatar, translated where provided in the app’s languages. You can also read this notice before using the service.
The label belongs to the interface: it should not be treated as a marking embedded in every image, video or audio track that someone might record or share outside the service. Even when a recording does not show the label, the animated face, synthetic voice and responses remain AI-generated content.
Consent to use photographs and voice does not change this artificial nature. Anyone sharing content must follow applicable transparency rules and must not misleadingly present it as authentic recordings of the depicted person, as explained in the Terms.
The engine receives instructions to use available memories, acknowledge when it does not know a fact, avoid inventing names and dates, and not give medical advice. These are instructions to the model, not guarantees of results: they may not be followed in every response.
The avatar can make mistakes, confuse people or events, and produce inaccurate or inappropriate information. It is not a reliable source for establishing what actually happened. Responses do not replace advice from a doctor, psychologist or legal, tax or financial professional.
The avatar does not know the depicted person’s thoughts. Even when speaking in the first person or using a familiar tone, its sentences are generated by the system and are not new statements, intentions or feelings of that person.
The engine is instructed to direct someone reporting a possible danger to themselves or others towards a doctor or the authorities. It may nevertheless fail to recognise danger or respond inadequately.
PERSONLIS is not an emergency service and does not contact emergency responders for you. In an emergency in Italy call 112; elsewhere contact your local emergency number.
Duration depends on available minutes, with a technical limit of approximately one hour per session and only one simultaneous conversation per account. Minutes include avatar preparation and waiting during the session, as described in the Terms. When the service reports that minutes have run out, the call closes.
You can close the call at any time using “End conversation”. The service requires an Internet connection.
Separate choices are required to use the photograph and voice recording, as described in Avatar Consents. Anyone uploading another person’s materials must have that person’s demonstrable authorisation. Using minors’ photographs or voices for the avatar is prohibited. If use is disputed, the Provider may request evidence and consider suspending use during verification.
The photograph and recording are used to produce the animated face and synthetic voice. Questions, relevant memories, useful exchanges and visual information may be processed by external AI services to generate responses. Data processed, providers, storage locations and retention periods, and rights are described in Privacy and GDPR.
PERSONLIS is operated by LAPOFEIMH DI PEGNA FABIO LUIGI, Via Castellaro 25, 47843 Misano Adriatico (RN), Italy — VAT No. 04776290407.
For questions about the avatar or to report an inappropriate response, write to legal@personlis.com. For help using the service: help@personlis.com.