Archived copy. This is AI and Synthetic Content Policy, Version 1.0 — Effective 20 October 2026, exactly as Kavi published it. This page is never edited. SHA-256 of the source text: 37414ae35a6f6dea68179ed235014ece6bb89303582c001ca35e137e01aa4ef0. Current version: kavi.audio/legal/ai-synthetic-content-policy. All versions: kavi.audio/legal/policies.json.

AI and Synthetic Content Policy

Kavi Media, Inc.
Version 1.0 — Effective 20 October 2026
Published at kavi.audio/legal/ai-synthetic-content-policy

1. About this Policy

1.1 This Policy is one of the Kavi Creator Policies. It covers: (a) what you must disclose when your content includes AI-generated or AI-altered voice, music, narration or script; (b) how synthetic audio is labeled for listeners; (c) the limited ways Kavi itself uses AI on creators' content; and (d) the rule that a replica of your voice is made only under a separately signed AI-Voice & Digital Replica Consent or another signed agreement that specifically permits it (such as a Voice License Agreement).

1.2 This Policy is issued by Kavi Media, Inc., a Delaware corporation with its principal place of business in Alpharetta, Georgia, USA (mailing address: 169 Madison Ave STE 30314, New York, NY 10016, USA) ("Kavi").

1.3 In this Policy:

2. Who this Policy applies to

This Policy applies to every creator, co-host, co-owner, author and approved short-form creator with a Kavi Creator Portal ("Portal") account, for all content they deliver or post to Kavi.

3. Using AI tools in your own content

3.1 AI tools are allowed. You may use AI tools to make your content, as long as you follow this Policy, the Community Guidelines and your signed agreement. Kavi does not reject content because of the tools you used, only because it is undisclosed, unlabeled, unlawful or otherwise breaks the Creator Policies.

3.2 The tool's terms must allow it. Use only AI tools whose terms allow commercial use of the output in the way you license it to Kavi.

3.3 No synthetic voice of a real person without consent. Do not include a synthetic, cloned or imitated voice or likeness of any real, identifiable person (living or dead) unless that person has given you written consent covering this use. After the person's death, you need written consent covering this use from the person or persons then legally entitled to control the applicable postmortem voice, likeness, publicity or digital-replica right, including any transferee, successor in interest or other person authorized by applicable law. An estate or legal representative may give that consent only to the extent it holds that authority, and Kavi may require reasonable evidence of that authority. You also need written consent before including a synthetic voice of any co-host, co-owner, guest or cast member. Keep the consent and give it to Kavi on request (and, for shows, attach it to your AI disclosure).

3.4 Music. (a) A music release whose sound recording is wholly generated by AI, with no human performance and no human creative selection, arrangement or editing of the sound, may be delivered only with Kavi's prior written approval. (b) No release may imitate the voice or likeness of a real person without that person's written consent.

3.5 No deceptive synthetic content. Synthetic content must not be used to deceive listeners about who is speaking, to fabricate statements by real people, or to impersonate anyone. See the Community Guidelines.

4. Disclosing synthetic content to Kavi

4.1 What to disclose. When you deliver or post content, disclose any part of it that was generated or materially altered by AI, including AI-generated or AI-altered voices, vocals, narration, music, sound effects, lyrics, script text and artwork, and the tools you used.

4.2 What you do not need to disclose. (a) Technical processing that does not generate new expressive content, such as noise reduction, leveling, loudness processing and transcription. (b) For music, AI tools you used only for ideas or reference, where no AI-generated sound or lyric remains in the release.

4.3 Where to disclose. (a) Shows (Creator Master Agreement): in the AI disclosure form delivered with each episode (state "none" if there is none), giving the time-code or section, type, tool used, a description of the AI involvement, the consent of any person whose voice is imitated, and the label applied. (b) Music (Music Master Agreement): in the Release Schedule. (c) Audiobooks (Audiobook Master Agreement): in the Chain-of-Title Certificate (Exhibit C3, item 10) for any part of the text, and, for any cover art you supply that was generated or materially altered by AI, in item 7 of that Certificate or by telling Kavi in writing when you supply it. (AI narration that Kavi produces is covered in section 8.) (d) Short-form: in the upload flow.

4.4 If you are resident in India, this disclosure is also your declaration, at delivery, of any synthetically generated content in your work, which Kavi needs in order to label it as India's IT Rules require.

4.5 Consequences. Kavi may reject, under your Master's acceptance process, content that contains undisclosed AI-generated material, and may remove or limit short-form items that do not comply. Kavi may add a label where synthetic content was not labeled.

5. How synthetic audio is labeled

5.1 Labels on every synthetic audio item. Synthetic audio on Kavi carries, at least: (a) where applicable law requires it, an audible disclosure that the item contains an AI-generated or AI-altered voice or audio, presented as that law requires; (b) a visible label on the item's listing; and (c) permanent metadata, including a unique identifier, identifying the content as synthetic. Kavi applies these labels in every country. Where India's Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, as amended in 2026, or any other law requires more for an item, Kavi adds what that law requires.

5.2 Kavi adds the labels. Kavi adds these labels, based on your disclosure, to content you deliver under a Master and to short-form items, and adds them to Kavi-made versions (such as AI-assisted dubbed versions).

5.3 Do not remove them. You must not remove, obscure or contradict any label, including in any version you later distribute elsewhere where your signed agreement allows that (for example, a dubbed version Kavi lets you redistribute, or an AI-narrated audiobook under its file license, must keep its synthetic-audio label).

5.4 Listener labels Kavi uses. Depending on the content, listeners see labels such as (and, for AI-narrated audiobooks, also hear):

5.5 Kavi will never describe AI-narrated content as narrated by a human or by a "professional voice", and will never present a replica voice as a live human performance.

6. How Kavi uses AI on your content

6.1 A closed list. Kavi uses AI and machine-learning tools on the content you license or post only for: (a) transcription; (b) captions; (c) search; (d) recommendations; (e) content moderation (including safety screening); (f) fraud detection; (g) detecting and labeling synthetic audio; (h) chaptering, metadata, and choosing promotional clips and Bites, and detecting natural pauses for advertising, in each case without generating or altering any person's voice; (i) audio processing for editing support you have elected (section 6.2); (j) translation or dubbing, only as your signed agreement and any AI-Voice Consent allow (see section 7); and (k) training, fine-tuning, evaluating and improving Kavi's own AI models, as section 6.5A describes (never music). Your signed agreement sets the exact list that applies to your content; if it lists fewer purposes, only those apply. Kavi does not use AI on your content for any other purpose, except for audiobook narration that Kavi produces under the Audiobook Master Agreement (section 8).

6.2 Editing support. If you elect Kavi's editing support for a show, Kavi and its editors may also use AI tools on your raw material only to provide that editing, for transcription, captions, search, noise reduction, loudness processing, removal of clicks, long pauses, false starts and filler, content moderation and fraud detection, as the Editing Support Agreement lists. Editors never use AI to generate, clone or regenerate any person's voice, except the voice of you, a co-host or a co-owner who has signed their own AI-Voice Consent covering that use, or of a guest or cast member under the separate consent described in their Guest / Participant Release or Cast & Writer Release (Editing Support Agreement clause 9.4).

6.3 Music. Kavi never edits, remixes or alters a music master, and never translates, dubs or re-voices music or lyrics, without your separate written consent. Kavi may transcribe and index lyrics internally, on systems located in the United States, only to support search and recommendations, and does not display them to listeners without a separate lyrics license. Kavi never uses music (masters, stems, vocals or lyrics) to train, fine-tune, evaluate or build any AI model, including any voice model, and does not let anyone else do so; no opt-in is offered for music (Music Master Agreement clause 5.6).

6.4 No model of your voice or likeness. Kavi does not use your content in or with any model to reproduce or imitate any person's likeness. Kavi does not use your content in or with any model to reproduce or imitate your voice, or the voice of any co-host or co-owner, unless that person has signed their own AI-Voice Consent covering that use. The AI-Voice Consent never covers the voice of any guest or cast member; their voices may be used in that way only under the separate consent described in the Guest / Participant Release or the Cast & Writer Release.

6.5 Vendors may not train on your content. Each AI vendor Kavi uses on creators' content is bound by its written terms, together with the opt-out Kavi has exercised under them, not to train, fine-tune, improve or evaluate any model on that content, other than Kavi's own models that a vendor trains solely for Kavi (section 6.5A). Kavi lists its current AI voice, narration and dubbing vendors, and the other AI tool providers that process creators' content, in the Approved AI Voice Vendor List, published at kavi.audio/legal/ai-voice-vendors and in the Portal, and uses them only under these limits. Dubbed versions use stock synthetic voices or a Kavi Voice only (section 9.9): Kavi never uses a voice-preserving or voice-matching feature of any tool, except for a person's own voice under an AI-Voice Consent that person has signed.

6.5A Kavi's own AI models. Kavi may use content you deliver under a Creator Master, Audiobook Master or the Short-form Terms, including its audio (such as tone, rhythm and intonation) and its transcript, free of charge, to train, fine-tune, evaluate and improve AI models that Kavi owns or exclusively controls, including speech-recognition, transcription, translation and speech-synthesis (text-to-speech) models. Kavi owns those models and may license them (speech-synthesis models only as a hosted service, never as a copy of the model), but never your content or any dataset containing it. No such model is ever used to imitate or reproduce your voice or anyone else's, and Kavi uses reasonable measures to stop that happening. If you live in India, Canada, Illinois, Texas or Washington, Kavi does this only if you tick the consent box when you sign. Everyone else may opt out. To opt out or withdraw, email privacy@kavi.audio with your name and "AI training opt-out" at any time (any clear request counts): within 30 days after receiving it, Kavi stops using your content in new training and deletes it from its training datasets, but models already trained are not retrained. Kavi never uses for this the voice or words of anyone under 18 (or under the age of majority where they live, if higher), or the voice of a guest, cast member or co-host who has opted out or has not consented. It never applies to music, including music delivered under the Music Master and any excerpt of it in a short-form item. This applies only to agreements and releases signed on version 1.0 (20 October 2026) or later.

6.6 No third-party training without your opt-in. Apart from Kavi's own models under section 6.5A (and never for music, section 6.3), Kavi does not license, provide or permit any third party (including Kavi's own vendors) to use your content to train, fine-tune or evaluate any AI model, except under a separate written opt-in that you sign for each work concerned. Under an opt-in: (a) you may revoke it at any time for future uses; (b) you receive 50% of the net proceeds Kavi receives for that use; and (c) if you are resident in India, from 13 May 2027 the opt-in will meet the consent requirements of India's Digital Personal Data Protection Act, 2023, and will be available in English and Telugu. Declining an opt-in has no effect on your agreement.

6.7 Scraping. Kavi uses reasonable technical measures against unauthorized scraping and bulk downloading of content from the Service. Kavi may, but need not, act against scraping or AI training of your content by others, with your cooperation.

7. Translation and dubbing

7.1 Shows. Under the Creator Master Agreement, Kavi may make translated, subtitled or dubbed versions of your episodes using human translators and voice artists or AI-assisted tools with stock synthetic voices or a Kavi Voice (section 9.9), none of which replicates your voice (never a voice-preserving or voice-matching feature, except under an AI-Voice Consent you have signed). Kavi gives you at least 14 days' written notice of each new language, and you may object to a language within that period, giving reasons; if you do, Kavi will not produce that language version of your show. A translated or dubbed version earns royalties at the same rate as the original. If you are resident in India or Canada, your rider may set additional conditions.

7.2 Short-form. Kavi publishes a dubbed or translated audio version of a short-form item only if you switch on "Allow dubbed versions" in the Portal, for the languages you choose. You may switch it off at any time.

7.3 Music and audiobooks. Translation or dubbing of music, or translation of an audiobook, happens only with your separate written consent.

7.4 Your own voice in a dub is used only under an AI-Voice Consent (section 9). Otherwise, a dubbed version uses a stock synthetic voice, a Kavi Voice (section 9.9) or a human voice, none of which is a replica of yours, and Kavi never uses a voice-preserving or voice-matching feature on it.

7.5 Every AI-assisted dubbed version is labeled as section 5 describes.

8. AI narration of audiobooks

8.1 Under the Audiobook Master Agreement, Kavi produces AI or hybrid narration of a book only with the express AI Narration Consent separately initialed in the Book Schedule by the licensor, by each living author other than the licensor (or, for a deceased author, by the person entitled to the author's moral rights) and by each co-owner of the title. A general signature of the Book Schedule is not an AI Narration Consent.

8.2 Kavi uses human narration for every title it designates a Flagship Title.

8.3 Kavi uses only text-to-speech vendors whose terms license audiobook production and distribution of the output, bar the vendor (together with the opt-out Kavi has exercised under them) from training on the book or the audiobook, and confirm that every human voice behind the synthetic voice was used with informed consent and is not designed to imitate the author or any other identifiable person. A Kavi Voice (section 9.9) may also narrate: it is made with the voice actor's consent under a Voice License Agreement, is designed to sound only like that actor, and is never the author's voice. Those vendors are named in the Approved AI Voice Vendor List (kavi.audio/legal/ai-voice-vendors).

8.4 The author's own voice is never cloned for narration except as Author Replica Narration under section 9.

8.5 Every AI-narrated audiobook carries the "Digitally narrated" label described in section 5.4.

9.1 Separate and optional. Kavi makes a digital replica or voice model of a creator's, musician's or author's voice only under a separately signed AI-Voice & Digital Replica Consent, and only for the Category and uses ticked and initialed there, or under a separately signed Voice License Agreement (section 9.9). Signing it is never a condition of any Master, Schedule, tier, rate, placement or payment, and declining or revoking it has no effect on them. Each co-host, co-owner, featured artist and co-author signs their own; guests', cast members', session musicians' and narrators' voices are never covered.

9.2 The Categories. (a) Category (i), creators under the Creator Master Agreement: dubbing of episodes of the shows listed, in the languages ticked, with your replica voicing only your own lines, translated faithfully; and, if ticked, short pickups to correct mistakes or remove material for legal or content-policy reasons, within the limits and notice period in the Consent. (b) Category (ii), musicians: Kavi will never use your replica to create singing, rapping, humming, chanting or any other musical vocal performance, any new musical recording or any change to a master. If ticked, only short pickups in your spoken-word intros for the releases listed. (c) Category (iii), authors: if ticked for a title, narration of that title only (your own work) with your replica voice, and third-party distribution only if separately ticked.

9.3 Never allowed. Whatever is ticked, Kavi will never use your replica voice or voice model for advertising, sponsorship reads or endorsements; political, electoral or advocacy content; sexual content; defamatory, hateful or harassing content or anything that misrepresents your views; any content other than the covered works; live, real-time, interactive or conversational uses; audiobook narration (other than your own covered titles under Category (iii)); musical vocal performance; or sharing the voice model or your voice data with anyone other than approved vendors acting for Kavi.

9.4 Your approval. Before Kavi first publishes replica output in a use you ticked, it sends you samples. If you do not respond within the approval period, the samples are treated as rejected. Kavi publishes nothing in that use until you approve.

9.5 Revocation and deletion. You may revoke the AI-Voice Consent, fully or for any Category or use, by written notice (including by email to privacy@kavi.audio or through the Portal). Revocation takes effect after the 30-day Revocation Notice Period in the AI-Voice Consent. If you live in Washington State, USA, or you are resident in India and your notice says you withdraw consent under the Digital Personal Data Protection Act, 2023, Kavi stops creating replica output when it receives your notice, and the 30-day deletion period runs from that date. Kavi then deletes the voice model and voice data from its own systems, and instructs its vendors to delete them, within 30 days, and confirms deletion in writing; backup copies are deleted or overwritten in the holder's normal backup cycle and never restored or used.

9.6 Royalties. Replica outputs earn royalties at the same rate as the human-voiced equivalent, with no per-use fee and no deduction for the cost of the voice model, and are shown on separate statement lines.

9.7 Minors. The AI-Voice Consent is never available to anyone under the age of majority where they live.

9.8 The AI-Voice Consent itself sets out the full terms. On anything about digital replicas in your own work, the Consent controls.

9.9 Kavi Voice (Voice License Agreement). Kavi may pay an adult voice actor to license their voice under a separate Voice License Agreement, so that Kavi can build its own synthetic voice (a "Kavi Voice") for the uses specifically described in that agreement, such as audiobook narration, dubbing and Kavi's own programs and promotion. A Voice License Agreement covers only the voice actor who signs it. A Kavi Voice carries the voice actor's own name, with the words "Digitally narrated" shown somewhere on the Service. It never uses your content and never imitates your voice, and the actor can withdraw it at any time.

10. Contact

11. Changes to this Policy

11.1 Kavi changes this Policy only as the Policy Acknowledgements describe: (a) at least 14 days' notice by email to the address in your Portal account and by a notice in the Portal, with a summary of what changed; (b) shorter notice, or immediate effect, only where required by law, a court or regulator order, an app store or payment provider, or needed to deal with an urgent safety, security, fraud or child-safety risk (for example, a change to labeling rules under India's IT Rules), and only to the extent that requirement or risk demands, with an explanation as soon as practicable. Shorter notice affects only when the change takes effect, and paragraph (f) applies to it in full; (c) a change in your favor, or a purely administrative change, may take effect when posted, with notice; (d) a change that expands what Kavi may do with your content or personal data, including any new AI use, is a material change: Kavi will ask you to acknowledge this Policy again in the Portal. Until you acknowledge, the earlier version keeps applying to content under any Schedule, Release Schedule or Book Schedule signed before the change, as your Master provides, and Kavi will not hold back uploads of that content, royalties already earned, or exploitation of or payment for content already delivered. Kavi may ask you to acknowledge the new version before you sign a new Schedule; (e) if you do not accept a material change, your Master says what you may do; and (f) a change does not apply to content delivered, royalties earned or disputes that arose before it took effect, and does not apply as a new acceptance criterion or an increased production or delivery burden to content under a Schedule, Release Schedule or Book Schedule signed before the change, except to the extent applicable law requires immediate application. A court or regulator order is implemented to the extent it legally binds Kavi. An app store or payment provider requirement, or an urgent safety, security, fraud or child-safety risk, may justify only action that your signed agreement or applicable law already authorizes (for example, a mandatory takedown); it does not itself amend your signed agreement or make a new obligation in this Policy apply to earlier content. Royalties already earned on removed or restricted content remain payable.

11.2 No change to this Policy can expand the AI uses your signed agreement allows, reduce your royalty rates, change your payment terms, or change the rights you have granted or kept under your signed agreement. Any new AI use of your content beyond your signed agreement needs a written amendment or consent signed by you.

11.3 Every past version stays available in the Portal.

12. Language

This Policy is in English, and the English text is binding. Kavi may provide a Telugu translation or summary as a courtesy; it is for information only. If you are resident in India, the AI-Voice Consent and any AI-training opt-in will be available in English and in Telugu, and on request in any other language listed in the Eighth Schedule to the Constitution of India, by 13 May 2027.

13. Your signed agreement prevails

If this Policy conflicts with your signed agreement, including your AI-Voice & Digital Replica Consent, your signed agreement prevails.