Editorial & AI Policy

Effective: June 6, 2026 (v1.0)

Demades episodes are AI-generated: scripts are written by AI from research, and the hosts you hear are synthetic voices, not real people. This page explains the editorial rules our generation pipeline follows for content about real companies and people, and how to request a correction or takedown.

AI disclosure

Demades is an AI-native production house, and you choose whose voice is on the mic. We disclose what is actually AI, and only that:

• Generated episodes (AI hosts voice the whole episode) carry an AI-disclosure notice on the episode page and a short spoken note in the audio that the episode was generated with AI by Demades.
• Recorded episodes (your own real human recording, which we research-pack, edit, and produce) are not AI-voiced, so they carry no AI-generation disclosure — the voices are real. The AI here is the production work, not the voice.
• Hybrid episodes (real recorded segments plus AI-host segments) disclose the AI portion: the page notice and the spoken note apply to the AI-host segments, and show notes indicate which segments are recorded versus AI. The recorded segments are real.

Either way, companies and people discussed in an episode are not affiliated with and have not endorsed Demades or the episode unless expressly stated. The editorial rules below apply to the research and content of every episode, regardless of whose voice is on the mic.

Opinionated and unverified episodes

Some episodes are created from prompts that ask for a deliberately opinionated, one-sided, or unverified point of view. We allow these, and we label them clearly: a flagged episode shows a prominent warning on its page and includes a spoken note in the audio that it presents an opinionated or unverified point of view and should be treated as opinion, not established fact. Labeling an episode this way does not loosen the rules below — knowingly false *damaging* claims about a real person or company, harassment, and illegal content are refused regardless.

Editorial rules for real companies and people

1. Sourced or hedged. Factual claims about named companies or people must trace to a source gathered during the episode's research phase. Claims that cannot be sourced are hedged ("reportedly," "according to ...") or cut. 2. No unverifiable damaging claims. Reputation-damaging assertions that cannot be verified are removed, not hedged. 3. No commissioned attack content. We accept episodes *about* any company, including competitors of paying customers, under the same neutral editorial standard — but we refuse commissioned content whose purpose is to disparage a third party, and no customer can pay to shape our coverage of anyone else. 4. Trademarks and likeness. Company and product names are used only to refer to the companies themselves (nominative use); all trademarks remain the property of their respective owners, and we do not use a third party's logo or trade dress or imply any affiliation, endorsement, or sponsorship. We do not imply that a named person said or endorsed something they did not. 5. Comparative claims. Where an episode compares Demades to a named product, any claim about that product is truthful and substantiable from public information; we frame our own strengths rather than disparage, and we correct or remove a comparison that becomes inaccurate as a product changes (see the corrections path below). 6. No reproduction of copyrighted works. An episode is transformative commentary, analysis, review, or reporting about a subject — it is never a way to reproduce someone else's protected work. We decline prompts that ask us to read or narrate substantial copyrighted material (song lyrics, poems, passages from books, articles, or scripts, or paywalled content) or to recreate a specific copyrighted show, film, or its characters. An episode *about* a work is welcome; an episode that simply *reproduces* one is not. 7. Reference images and likeness. When you build an AI-video character, you may attach a reference image only if you have the rights and consent to use it. We ask you to confirm that explicitly and to say what it is — your own photo, original or licensed fictional art, or someone else's image you hold the rights to — and we record that confirmation with the upload. Uploads are reviewed before they can shape a generated video, and we decline images we cannot verify. We do not generate characters meant to depict a real person saying or doing something they did not: permission to use a picture is not permission to put words in someone's mouth.

Corrections and takedowns

If you believe an episode states something false or misleading about you or your organization, or uses your protected work or likeness improperly, email [email protected] with the episode link and the specific statement at issue.

• We acknowledge within 2 business days and aim to resolve within 5 business days.
• For credible defamation or right-of-publicity complaints, we unpublish the episode while we review.
• Outcomes: correction (regenerated segment), correction note on the episode page, or removal.

Copyright infringement claims (DMCA, 17 U.S.C. § 512) are governed by our DMCA & Copyright Policy; notices go to the same address.

Contact

D3SMC LLC (d/b/a Demades) · Austin, Texas · [email protected]