A human responsibility
TV Pater may employ AI tools to help prepare texts, organise materials, and create content. The responsibility for choosing what to publish remains with TV Pater Limited. The output of a model is not a historical source, a doctrinal certification, or a guarantee of truth. The site's searches query the published catalogue: they are not conversations with an artificial spiritual assistant.
Non-derogable rule on people
TV Pater does not clone or imitate the voice of real people, nor does it provide exceptions based on authorisation. It does not fabricate statements by priests, popes, saints, or other people. It does not use an artificial voice to make it seem that a person has spoken words they did not speak. Images and videos must also not simulate authentic testimonies or a non-existent personal approval.
Synthetic voices in podcasts
For podcasts produced with a generic synthetic voice, TV Pater provides the label “AI voice synthesis” in the app, corresponding to the individual podcast, near the title or the playback command. The indication must be legible, accessible, and visible before the start of listening, without having to open a separate legal page. The text and publication remain under the editorial responsibility of TV Pater.
This mode concerns the presentation of the podcast in the app. For other channels, any required information is assessed in relation to the content and the mode of use. No general editorial obligation is established to insert spoken announcements in audio files or radio programmes.
AI Act and contextual transparency
The European Artificial Intelligence Act distinguishes between the obligations of providers and those of deployers. Article 50 does not prescribe a spoken announcement for every use of voice synthesis. For audio, image, or video content that qualifies as a deepfake, it requires that its artificial nature be declared clearly and distinguishably, at the latest at the first exposure, and in a manner that respects accessibility. Qualification depends on the content and context: not every synthetic voice is automatically a deepfake.
The European Commission distinguishes between indications perceivable by the public and the technical markings required of providers. For texts of public interest, the specific exception linked to human review or editorial control with responsibility for publication does not constitute a general exemption for audio, images, and video. The presence of a label in the app does not, in itself, certify compliance with any obligations on other channels.
Pre-publication checks
The editorial team must record the tool, function, rights, person responsible for control, distribution channels, and the notice applied to the content. For tools that produce provenance markings, these must not be intentionally removed. Persons using AI for TV Pater must receive instructions appropriate to the activities and risks. The introduction of new interactive features or new data processing requires a check before activation; this document is not a general certification of compliance.
Data, dignity, and reports
Do not use AI tools to infer people’s faith, manipulate vulnerable individuals, or automate significant decisions regarding supporters. Do not input confidential user or donor data into generative tools without an appropriate legal basis, agreements, and controls. To report a false attribution, an undeclared voice, or another issue, write to specifying the content, channel, and, for audio, the minute of the passage.