AI-generated content disclosure

What is generated

Three parts of a rendered video are machine-generated:

• The narration audio. Synthesised from the script, either with a built-in voice or a voice cloned from reference audio you supplied, where you confirmed you own that voice or hold consent to use it.
• The presenter's mouth movement. Generated frame by frame to match the narration audio. Head pose, body and background come from the source clip you registered.
• The draft script. Produced by a language model from the text extracted from your slide. You review and can rewrite it before rendering.

The slide imagery itself is your own content, rendered from your file.

How it is produced

Rendering runs as a pipeline: the deck is rasterised page by page, a draft script is generated and made available for editing, audio is synthesised, and lip-sync is generated per page from the final audio. Pages are then encoded into a single video.

On self-hosted deployments every stage runs on hardware you control. On managed cloud, stages run on our infrastructure and are deleted according to our Privacy policy.

How output is marked

• A subtle watermark identifying the output as AI-generated is composited into the video.
• AI-generation metadata is written into the output file so downstream platforms and forensic tools can identify it.
• Enterprise customers may request provenance metadata conforming to C2PA where supported.

You may not remove or obscure these markers where removal would make the synthetic nature of the content materially less apparent.

Consent requirements

Before you register a presenter or clone a voice, you must be that person or hold their written consent. This is a condition of use, not a suggestion — see our Terms of service.

If your organisation is registering presenters on behalf of employees, keep the consent records. We may ask to see them.

Known limitations

• Pronunciation of uncommon proper nouns and technical terms may be wrong. Check the audio.
• Mouth movement is synthesised, and can look unnatural with very fast speech or heavily accented audio.
• The draft script reflects what the model inferred from your slide. It can state something your slide does not support, which is why script review exists.
• Lip-sync quality depends heavily on the quality of the source clip: frontal, evenly lit, no occlusion.

Reporting misuse

If you believe a video was produced using your likeness or voice without consent, email abuse@docpresenter.com with the video URL and proof of identity. We act on valid reports, remove the content, and terminate the account that produced it.