# InVideo AI for Faceless YouTube Videos: Workflow, Costs, Rules

> How InVideo AI builds faceless YouTube videos and Shorts: script, AI voice, b-roll, captions, credits by plan, and what invideo says about YouTube's rules.

_Source: https://professionalstoolkit.com/articles/invideo-ai-faceless-youtube-videos — The Professional's Toolkit · updated November 2026_

---

InVideo AI can build a faceless YouTube video from a script you supply: an AI voice reads it, stock or generated b-roll plays underneath, captions follow the words and music sits under the narration [1]. Editing and exporting cost nothing. The voice, the transcript behind the captions and any generated footage spend credits, and by default Agent Two generates without stopping to ask unless you change its approval setting. Starter, at $20 a seat a month billed monthly, can suit a channel built on stock footage and text to speech if the voice models you need appear in its model list, and it can generate light b-roll with Seedance 2.0 Fast or 2.0 Mini. Seedance 2.5 and full Seedance 2.0 start on Plus. Two limits frame the rest. invideo hands you a file and stops there, and its own blog warns that near-identical text to speech over stock video is what YouTube now treats as risky.


**400** — Credits a seat a month on Starter ($20 a seat a month billed monthly); they reset monthly and never roll over.



- **0 credits** — Timeline edits and exports; voice, transcripts, generation and paid stock cost credits
- **1080p** — Export ceiling for a finished timeline, 16:9 or 9:16
- **July 2026** — When YouTube renamed repetitious content to inauthentic, per invideo


## Two routes into the same faceless video

In November 2026 the default is **Agent Two**, invideo's conversational agent launched in August 2026, working on the invideo Editor timeline launched in September 2026. Its Script to Video and Explainer workflows run as a conversation, while Autopilot produces a draft in one shot with no back and forth. What comes out is not a locked render. It is a timeline of real clips you can recut, recaption or undo, and every change there is free. Starter seats get Agent Two Lite only, the fast tier that is light on credits; Pro and Ultra open from Plus upward.

The original InVideo AI flow — paste a script or prompt, get a stock-media video with AI voiceover and subtitles, then edit it with Magic Box commands — is still reachable through the Older versions button on the prompt page, while the default is now Agent Two, which can do the same in a guided conversation or in one shot with Autopilot. That older route is the one invideo's faceless generator page describes, drawing on 16 million-plus stock images and videos. Its subtitle styles include karaoke, word by word and Hormozi style, and an SRT file downloads after export. Avatars and AI twins live behind Older versions as well. Our [text to video guide](/articles/invideo-ai-text-to-video) compares both routes beyond the faceless use.

## From script to upload-ready file

invideo's faceless how-to starts from a script you already have, not from a topic. That order matters. The script is where a channel's originality lives, and the workflow assumes you bring it rather than asking the agent to invent one per video. Agent Two still helps tighten a draft, but the topic and the angle are yours. Put together from the docs and invideo's own pages, the steps look like this:

1. Paste the script and set the format: 16:9 for a long video, 9:16 for a Short.
2. Pick a narrator from the ElevenLabs TTS v4 voice library, or clone your own voice with Seed Audio 1.0 from a short, clean clip and tag your lines in the prompt.
3. Choose the b-roll: stock clips, generated shots, or your own footage mixed with generated scenes.
4. Build captions from a transcript, where each word carries its own start and end time.
5. Lay music under the voice; a ducker effect lowers it when the narrator speaks and lifts it in the gaps.
6. Review the cut, export a 1080p MP4 and upload it yourself.

A voice ID chosen in step 2 works only with the model that made it, so switching narration models means picking the voice again. Word-by-word captions stay an editable track on the timeline until export, which keeps typo fixes free. Cloning is where the rules tighten. InVideo AI requires explicit consent from the person in the recording, captured in a separate in-product step that can be withdrawn; celebrities and politicians cannot be cloned, and an avatar of anyone under 18 is forbidden even with parental consent. For a faceless channel the clean case is your own voice, and our [voice cloning guide](/articles/invideo-ai-voice-cloning) covers the setup.


> 💡 **Keep captions on the timeline:** Captions built from a transcript stay editable until export. Burned-in subtitle models write a new file, price by video length and must be run again for any change.


## What a faceless video costs in credits

Every timeline edit and every export is free, whoever makes it. Credits go on the narration, the transcript behind the captions, generated b-roll, paid stock placed on the timeline and Agent Two's own work. Outside agents working through invideo's MCP server run a dry run and wait for a yes on anything costing one credit or more, while calls under one credit run without asking. Agent Two's default is "Always allow generation"; switch it to ask before video or before any generation if you want to approve each paid step. As read in November 2026, ElevenLabs text to speech cost about 3.20 credits per 1,000 characters, and invideo reserves the right to change credit costs at any time without notice.

| Plan | Billed monthly | Credits a seat a month | Agent Two tiers | Seedance 2.5 | Premium stock allowance a month |
|---|---|---|---|---|---|
| Starter | $20 a seat a month | 400 | Lite only | No (2.0 Fast and 2.0 Mini only) | 100 |
| Plus | $60 a seat a month | 2,000 | Lite, Pro, Ultra | Yes, at 1080p | 200 |
| Max (5,000 credits) | $150 a seat a month | 5,000 | Lite, Pro, Ultra | Yes | 500 |

The annual card showed Starter at $10 a seat a month billed annually, an offer shown in November 2026. Under that offer an annual Starter, Plus or Max seat costs the same as six monthly payments, so annual billing only pays if the channel will run longer than six months, and the annual price is charged in full upfront. Plan credits reset monthly and never roll over. Fees are non-refundable: invideo's Terms allow a refund request only within 24 hours of a first purchase or renewal, annual included, and only if no credits or premium features were used, and credits, including extra credits, are never refunded; the [pricing breakdown](/articles/invideo-ai-pricing) has the full rules. The Free plan is the way to try the workflow first. It keeps the Editor and its exports free, but credit-based AI actions sit under a weekly limit whose size is not published.

The plan choice follows the b-roll. A channel built on stock footage and text to speech can fit Starter if its model list carries the voices you need, because generated video, the costly ingredient, barely enters the recipe; invideo estimates Starter's credits at about 15 Seedance 2.0 Fast videos a month. Generated shots change the math fast. Video models charge per second, and at about 96 credits per five-second 1080p Seedance 2.5 clip, Plus covers about 20 clips a month and Max about 52, by our arithmetic, before any retries. Retries are not rare: invideo's own documented productions average about three generations per usable shot. Going past the monthly premium stock allowance also triggers a charge at download.

## Running a series without running copies

For volume, invideo documents a bulk recipe built around sleep and meditation videos. You set up one timeline, duplicate it for each variant and swap the picture and sound, so many channel videos come from a single setup [2]. Rain or wave beds come from a looped ElevenLabs sound-effects model, since the docs call a music model a poor fit for rain. Very long videos are exported in parts and joined into one long MP4 with a stitch operation, each part set by its own start and end in seconds.

invideo's blog defines YouTube automation as automating ideation, scripting, creation and management so a channel runs with minimal human involvement. Its own automation page covers less than that. It supports a series through a reusable brief that holds audience, format and preferences, and it says each video should be reviewed before upload [3]. **invideo does not manage or publish to your channel.** You export the file and upload it, which also means titles, thumbnails and scheduling stay with you.

Shorts come from the same project. Each platform version sits on its own branch and exports at 1080x1920, and safe-zone guides keep captions clear of app buttons without appearing in the file. Reframing a 16:9 shot by hand is free; generative reframe models paint new picture, charge by source length and do not track a face. The [Reels and TikTok guide](/articles/invideo-ai-reels-and-tiktok) covers the vertical side in depth.


> 💡 **Branch Shorts, do not rebuild:** Put each vertical 9:16 version on its own branch of the same project, so the long video and its Shorts share one setup.


## YouTube's inauthentic-content rule, in invideo's words

invideo's blog says YouTube renamed repetitious content to **inauthentic content** in the Partner Program in July 2026. It adds that YouTube judges the channel as a whole rather than each upload: the theme, top and recent videos, the share of watch time, metadata and the About section [4]. A single good video does not rescue a channel whose catalog looks mass-produced.

The same post sorts formats by risk. invideo lists Reddit narration, script and thumbnail recycling farms and ambient loops as the formats most exposed to demonetization. It places AI used for editing speedups, visuals and cloning your own voice in the safe zone, and calls near-identical text to speech over stock video risky. Notice the tension. The bulk sleep recipe above is an ambient-loop format, and duplicating one timeline while swapping the picture is exactly the pattern the blog flags.

The practical reading is to let the tool speed up production while the channel's substance stays yours: original scripts, varied structure, your own cloned voice where possible, and a human review of every upload. invideo's pages say AI-generated videos are not automatically monetizable and that each platform sets its own rules, and the faceless how-to promises no views, subscribers or money. The Terms also make you responsible for disclosing AI-generated content where the law or recognized standards require it.


> ⚠ **Template channels are the risky path:** invideo's own blog says near-identical text to speech over stock video is risky under YouTube's inauthentic-content rule, and lists ambient loops and Reddit narration as highest risk.


## Who owns what you upload

The Terms grant a perpetual, irrevocable, worldwide license to use, modify, distribute and commercialize your outputs, and it survives cancellation. Two limits sit beside that grant. Outputs may not be unique and carry no exclusivity, so another channel can end up with similar generated or stock footage. Stock media stays under its own license. The help center does not settle whether YouTube can claim a stock music track, and invideo warns that copyrighted songs can get a montage muted, so stock or licensed music is the safer bed.


## FAQ

**Can ChatGPT or Claude build the video for me?**

They can work through the invideo MCP server, signing in with your own invideo account. Ordinary timeline edits use no invideo credits, generation does, and a ChatGPT plan does not include invideo.

**Can I monetize videos that use invideo's avatars?**

invideo's help center says videos made with its AI avatars can be monetized on any platform, but each platform applies its own rules and invideo guarantees no earnings. The avatar library sits behind the Older versions button.

**Does the Free plan watermark my videos?**

invideo says exports from the invideo Editor are free and carry no watermark on every plan, Free included. For videos made in the classic flow or with Agent Two on Free, the current help center does not say.

**What happens to my channel videos if I cancel?**

The account falls back to Free at the end of the paid period, projects stay editable and export free, and videos made on a paid plan can still be used. Unused credits, including purchased extra credits, expire.


## The bottom line

InVideo AI handles the production half of a faceless channel well: script in, narrated and captioned 1080p video out, with free edits and a price you can approve before paid steps once Agent Two is set to ask before generating (its default is to generate without asking). Starter covers stock-and-voice videos and light generation with Seedance 2.0 Fast or Mini; Seedance 2.5 b-roll needs Plus or a higher plan, plus a retry budget. What it will not do is run the channel. Publishing, scheduling and the originality YouTube looks for stay with you, and invideo's own blog is clear that cloned-template uploads are the risky path.



## Sources

[1] invideo docs: Quickest path — https://docs.invideo.io/how-to/creators/faceless-youtube-video (2026-11)
[2] invideo docs: Quickest path — https://docs.invideo.io/how-to/creators/sleep-and-meditation-videos-in-bulk (2026-11)
[3] invideo: youtube automation — https://invideo.io/make/youtube-automation/ (2026-11)
[4] invideo blog: youtube kills ai faceless channels — https://invideo.io/blog/youtube-kills-ai-faceless-channels/ (2026-11)
[5] invideo pricing (US) — https://invideo.io/pricing/
[6] invideo Help: subscriptions and refunds — https://invideo.io/help/billing/subscriptions-and-refunds/
[7] invideo Terms of Service — https://invideo.io/terms-of-service/
[8] invideo news — https://invideo.io/news/