How to Make Faceless YouTube Videos with AI: 7-Step Workflow and Cost per Minute (2026)

To make a faceless YouTube video with AI, pick a niche, write a script, turn it into a voice-over with text-to-speech, cover the narration with AI images and video clips, add captions and music in an editor, and finish with a 16:9 thumbnail. On Kenerate, a finished minute costs about 114 credits with AI stills only, about 325 with a stills-and-video mix, and about 1,850 if every second is AI video (prices checked 23 October 2026). To keep the channel monetisable, each video needs real narrative, commentary or educational value, and you must disclose realistic AI scenes when you upload.

This is the production workflow, step by step. If you're still choosing a topic, start with our faceless AI YouTube channel ideas. If you're making vertical Shorts rather than long-form videos, the AI YouTube Shorts workflow is the better guide.

What is a faceless YouTube video?

A faceless video is one where the creator never appears on camera. The video is carried by narration and visuals instead: b-roll, illustrations, animation, maps, screen recordings or an animated host. Common formats include history and science explainers, story channels, business breakdowns, tutorials and ambient music.

AI now covers almost every part of the job: the script draft, the voice, the visuals, the captions and the thumbnail. What it doesn't replace is the part YouTube pays for: a point of view, research and editing choices that make each video worth watching.

What do you need to make faceless YouTube videos with AI?

Credit packs are one-time purchases, from $15 for 1,300 credits to $109 for 10,900, so a credit costs about one US cent. Video generation needs a paid pack. New accounts get 25 credits to try the voice and image tools.

How do you make a faceless YouTube video, step by step?

Step 1: Pick a niche you can make 50 videos about

Faceless channels grow on volume and consistency, so choose a subject with an endless supply of topics and a format you can repeat without the videos looking the same. Explainers, history, science, business stories and tutorials work well because each video has different substance. Our channel ideas guide compares nine niches.

Step 2: Write the script (this is where the value lives)

Use AI to help with research and outlines, but make the script yours. YouTube lists using AI for an outline or script as production help that doesn't need disclosure. What gets channels demonetised is a script with nothing in it.

A simple structure for an 8–12 minute explainer:

1. Hook (first 15 seconds). The question the video answers, and why it's surprising.
2. Context. What the viewer needs to know first.
3. Three to five sections, each with one fact, one example and one visual idea.
4. Payoff. The answer, plus your take on it.
5. Next video. Point to a related video on your channel.

Narration runs at about 150 words per minute, so a 10-minute video needs about 1,500 words. Write a visual note next to each paragraph (for example, "map of Rome, 117 AD, slow zoom"). Those notes become your image and video prompts in Step 4.

Step 3: Generate the voice-over

Paste the script into Voice Studio, pick an engine and a voice, and generate. Keep the same voice for every video, because it's your channel's identity. A minute of narration is roughly 900 characters.

Generate long scripts in sections (one per chapter). That way a mispronounced word costs you one paragraph, not the whole take. If you want the channel to sound like you without recording, you can clone your own voice. YouTube lists cloning your own voice for voice-overs as a minor change that doesn't need disclosure. Never clone someone else's voice without their permission.

Step 4: Make the visuals

Go through the visual notes in your script and make one visual for every 5–6 seconds of narration. Mix three kinds:

• AI stills for most of the runtime. You add slow pans and zooms in the editor. Seedream 5.0 Flash costs 8.1 credits per image and GPT Image 2.5 costs 7.2 at medium quality (1K). If a recurring character or mascot has to look the same in every video, use Nano Banana 2.1 (up to 14 reference images, 21 credits at 1K).
• AI video b-roll for the moments that need motion: the hook, the transitions and the payoff. Wan 3.0 makes 2–30 second clips at 15 credits per second at 480p, 30 at 720p and 60 at 1080p. For more cinematic shots, Seedance 2.0 costs 72 credits per second at 720p. Animating one of your stills (image-to-video) keeps the look consistent with the rest of the video. See which AI video generators YouTubers use for a model-by-model comparison.
• Real material where it's more honest than AI: screen recordings, charts, maps and licensed archive footage.

Make everything 16:9 for long-form videos. Draft at 480p and only re-render the clips you keep at 720p or 1080p.

Step 5: Edit, add music and captions

Put the voice-over on the timeline first, then cut visuals to it. A desktop editor such as DaVinci Resolve or CapCut handles a 10-minute timeline best. For short pieces, the Kenerate Agent can join 2–20 finished clips into one video of up to 3 minutes and mix in your voice-over and music, for a few credits (3 credits for two clips, plus 1 for each extra clip).

For captions, you have two options:

• Closed captions. Viewers can switch them on and off. YouTube can generate them automatically, or you can upload your own subtitle file. This is usually the right choice for long-form videos.
• Burned-in captions. These are part of the picture. Auto Subtitles transcribes the speech and burns it in with one of 10 caption styles. It costs 20 credits per clip up to 1 minute, 40 up to 2 minutes and 80 up to the 4-minute limit, and double at 1080p. It works best for Shorts, trailers and clips that people watch on mute.

For music, use YouTube's Audio Library or a track you have a licence for, or generate a short instrumental loop. Keep the music well under the narration.

Step 6: Make the thumbnail

The thumbnail decides whether anyone clicks. Use Image Studio's Thumbnail preset (16:9, four variants per run): one subject, three to five words of big text, high contrast. GPT Image 2.5 handles text best. Four variants at medium quality cost about 29 credits. Our AI thumbnail generator page has prompt examples.

YouTube's thumbnail rules: use 16:9 for videos, JPG or PNG, at least 640 pixels wide, up to 3840 × 2160. Custom thumbnails need a verified account. Upscale your favourite variant before uploading. Kenerate's image upscaler costs 3 credits.

Step 7: Upload and disclose correctly

In YouTube Studio, write a title and description that match what the video actually delivers. Then answer the altered or synthetic content question honestly (rules below).

How much does a faceless YouTube video cost per minute?

Here's what one finished minute costs on Kenerate, using the prices above and about 900 characters of narration per minute. These figures cover one clean pass, so budget extra for rerolls.

For a 10-minute video, that works out to about 1,140 credits on the budget recipe and about 3,250 on the balanced one. Add one thumbnail run (~29) and burned-in captions if you want them. Most channels land on the balanced recipe: stills carry the narration, and a few seconds of motion per minute keep it from looking like a slideshow. As shown below, a pure slideshow is also a monetisation risk.

Can faceless AI YouTube channels be monetised?

Yes, if the videos are original and add value. YouTube's monetisation rules don't ban AI or faceless videos. They ban a particular kind of video. Since 15 July 2025, the policy formerly called "repetitious content" has been called inauthentic content. It covers content that is mass-produced or repetitive, and it applies to the whole channel, not just one video. YouTube's examples include:

• videos that look like they're made from a template with little variation;
• image slideshows or scrolling text with minimal or no narrative, commentary or educational value;
• AI-generated content made with generic templates that gives the impression of mass production.

The reused content policy is separate. It covers channels that repurpose other people's videos with minimal changes, even with permission.

YouTube also won't monetise channels that use AI personas to give information on sensitive topics, such as an AI "doctor," "financial advisor" or "lawyer." A faceless finance channel narrated as you is a different thing from an AI character presented as a qualified advisor.

What keeps a faceless channel safe:

• Every video says something different: new research, a new argument, a new story.
• Your narration adds commentary or teaching, not just captions read over pictures.
• Visuals change with the content instead of looping the same template.
• You don't re-upload other people's clips, even with credit.

Do you have to disclose AI content on YouTube?

You must disclose content that is realistic and made or changed with AI in a way viewers could mistake for real. YouTube's help page lists these cases:

• a real person appearing to say or do something they didn't do;
• altered footage of a real event or place;
• a realistic scene that didn't actually happen;
• AI-generated music that's the main focus of the video.

You don't need to disclose AI used for scripts and outlines, captions, colour and lighting, beauty filters, clearly unrealistic content (YouTube's example is a person riding a unicorn), or cloning your own voice for voice-overs. The page doesn't say either way about a fully synthetic narrator voice. If the voice could be taken for a real, named person, disclose it.

Disclosure doesn't limit a video's reach or its monetisation, according to YouTube. Creators who repeatedly fail to disclose can have a label added by YouTube, have content removed, or be suspended from the YouTube Partner Program. For a history channel with photoreal AI "footage" of past events, tick the box. For a stylised illustrated explainer, you usually don't need to.

What mistakes get faceless AI videos ignored or demonetised?

• Slideshow syndrome. Static images and a monotone voice for 10 minutes. Vary the shots and add motion where it matters.
• Generic scripts. A summary anyone could produce in one prompt has no reason to exist. Add research, opinions and structure.
• Voice changes. A different voice every video breaks the channel's identity.
• Misleading realism. Photoreal AI scenes of real events without disclosure.
• Copied formats. Rewriting other creators' videos line by line drifts into reused content.

Sources

• YouTube Help: YouTube channel monetization policies (inauthentic content, reused content, AI personas on sensitive topics; 15 July 2025 update)
• YouTube Help: Disclosing use of altered or synthetic content
• YouTube Help: Add video thumbnails on YouTube (thumbnail size, format, verification)
• Kenerate Voice Studio, Image Studio, Video Studio and Auto Subtitles prices, checked 23 October 2026