Every figure on this page links to the organization that produced it, and each was re-checked against that source before publishing. What the faceless channels the how-to guides hold up actually post (our own measurement of 126 channels), what YouTube's monetization and AI-label rules say and when they changed, how many channels YouTube removes, how much of YouTube is AI-generated by anyone's count, how creators use AI voices and stock footage, and where the popular faceless-channel numbers came from.
The faceless channels the how-to guides hold up are old: the median one of 126 was created in 2016, and only 5 were started in 2023 or later. (Our measurement)See details
They post long videos: Shorts were 1.1% of the median active channel's uploads in the year to October 2026, and 50 of 111 active channels posted none. (Our measurement)See details
No collapse after the July 2025 rename: the same channels posted 8,227 uploads in the first half of 2025 and 7,604 in the first half of 2026, 7.6% fewer. (Our measurement)See details
On July 15, 2025 YouTube renamed its repetitious content policy to inauthentic content and said such content had always been ineligible for monetization; the policy did not mention AI until 2026. (YouTube)See details
From February 1, 2027, new YouTube Partner Program applicants will need 8,000 watch hours in a year, double today's 4,000. (YouTube)See details
YouTube removed 2,732,395 channels in April to June 2026, 2,310,181 of them for spam, deceptive practices and scams; it publishes no count of demonetized faceless or AI channels. (Google Transparency Report)See details
The public YouTube Data API returned YouTube's AI disclosure field on 0 of 15,509 uploads we read, so outside counts of AI video rely on their own judgement, not YouTube's label. (Our measurement)See details
The 38% and 40% faceless-share figures have no measured original, and the $10,000-a-month figure traces to a course seller's self-reported student results. (Origin traces)See details
What do the faceless channels the guides hold up actually look like?
Old, large and slow-posting. We took every channel that eight public "faceless YouTube channel" guides link by name as an example, 126 channels in all, and read them through the YouTube Data API on October 31, 2026. The median one was created in 2016 and had 1,695,000 subscribers. Only 5 of the 126 were started in 2023 or later. 79 had at least a million subscribers and 12 had at least 10 million. The guides agree on very little: only 21 of the channels appear in two or more of them.
The guides come from creator-tool and marketing publishers (Vidpros, Overseeros, AIR Media-Tech, Murf, InVideo, Descript, SubSub and Trends.vc), several of which sell tools for making faceless video. We did not judge whether each channel is truly faceless; a few show a presenter. So this is a read of the examples the guides chose, not a random sample of faceless channels. The per-channel and per-video data is published in channels.csv and videos.csv.
How often do they post, and how much of it is Shorts?
About once a week, and mostly long videos. The median channel that posted at all in the 365 days to October 30, 2026 put up 1.02 uploads a week. In that median channel, Shorts were just 1.1% of uploads, and 50 of the 111 active channels posted no Shorts at all. Pooled, Shorts were 24.4% of the 15,509 uploads, because a few very high-volume channels post many of them: one channel, WatchMojo, accounted for 18.9% of every upload in the sample. The median channel's typical regular video ran 16.35 minutes.
The typical regular video on the median channel had 60,236 views when we read it, against 29,393 for the typical Short. On only 13 of the 60 channels that posted both did the typical Short out-draw the typical long video, the reverse of what we measured for pro sports teams on our sports video statistics page.
Are the example channels still posting?
Most are. 15 of the 126 channels published no public upload in the 365 days we read, and 22 had nothing new in the last 90 days. 6 of the channel links in the guides no longer resolve to a channel at all; the API does not say whether those channels were renamed, deleted or removed.
There is no sign of a collapse after YouTube's July 2025 policy rename. The same channels published 8,227 uploads from January to June 2025, before the change, and 7,604 from January to June 2026, a fall of 7.6%. 59 channels posted less and 52 posted more. Uploads deleted since early 2025 are not in the count, which if anything understates the earlier half.
What else the public data shows
7.3% of the uploads carried captions the creator uploaded, 4.8% carried YouTube's paid-promotion label, and 0.8% were set as made for kids. The public YouTube Data API returned the altered-or-synthetic-content disclosure field on 0 of the 15,509 uploads, so nobody outside YouTube can count AI-labelled videos this way; anyone quoting a share of "AI videos" on YouTube is relying on their own classification, not YouTube's label.
Can a faceless channel be monetized on YouTube?
Yes. YouTube's rules never mention faces; they target content that is mass-produced, repetitive or reused. As of October 31, 2026, a channel joins the YouTube Partner Program's ad revenue share with 1,000 subscribers plus 4,000 qualified public watch hours in the last 12 months, or 1,000 subscribers plus 10 million qualified Shorts views in the last 90 days YouTube. A lower tier for fan funding and Shopping opens at 500 subscribers with three uploads in the last 90 days, plus 3,000 watch hours in 12 months or 3 million Shorts views in 90 days YouTube. YouTube pays partners 55% of net ad revenue on long-form videos YouTube and 45% of the Shorts ad revenue allocated to them YouTube. Applications are reviewed by automated systems and people who look at the channel as a whole, and a decision typically takes about 1 monthYouTube. YouTube may switch off monetization on a partner channel that has not posted for 6 months or more YouTube.
The bar rises in February 2027
On August 10, 2026 YouTube announced that from February 1, 2027, new applicants will need 8,000 qualified watch hours in the last 365 days, or 20M qualified Shorts views in 90 days, still with 1,000 subscribers; channels already in the program are not affected YouTube. To keep earning from the Shorts Creator Pool each month, a partner will need to hold 10M qualified Shorts views over the last 90 days; falling short does not remove a channel from the program or affect its long-form earnings YouTube. A partner channel will count as active with 1,000 qualified watch hours in a year, a million Shorts views in 90 days, or two long videos or five Shorts every 90 days YouTube. Existing partners have to accept the new terms by January 31, 2027 to keep fully monetizing YouTube. YouTube said the program has over 3 million creators and called this its first significant change since 2018 YouTube, and explained the higher Shorts bar by the more than 200 billion Shorts views it sees a day YouTube.
What did YouTube change in July 2025, and since?
A name, then the examples. On July 15, 2025 YouTube renamed its "repetitious content" monetization policy to "inauthentic content", to make clear it covers content that is repetitive or mass-produced YouTube. It said such content had "always been ineligible for monetization", and that its reused content policy for commentary, clips, compilations and reaction videos did not change YouTube. YouTube's policy-updates page announced the update for July 15, 2025YouTube; that notice first appears in the Internet Archive's July 3, 2025 capture of the page and not in the June 26 capture, so it went up about two weeks ahead of the date it names, July 15, 2025YouTube, archived.
Before the rename, the policy page described "Repetitious content" as a channel whose videos are so similar that viewers may have trouble telling them apart YouTube, archived July 10, 2025. The new wording, inauthentic content as "mass-produced or repetitive" content, first appears in the capture of July 11, 2025, four days before the stated date YouTube, archived. Two later changes came with no dated note on the page. Between May 31 and June 4, 2026 YouTube added "AI-generated content made with generic templates" as a named example YouTube, archived. The last capture showing the single section, "Inauthentic content refers to mass-produced or repetitive content", is dated July 9, 2026 YouTube, archived; by July 14, 2026 it had been split into three: "Generic or Repetitive Content", "Unsatisfying or Off-putting Content" and "AI Personas Related to Sensitive Topics" YouTube, archived.
What the current policy names as not monetizable
Not allowed to monetize (YouTube's wording, October 2026)
Source
"Image slideshows", templated storylines or scrolling text with little narrative, commentary or educational value
The same policy allows automated help. Its example of monetizable work is "using AI to edit your video scripts" or generating a unique background visual YouTube, and it says creators may use "automated tools or templates" as long as the final product shows their own effort YouTube. "Reused content" means repurposing material from YouTube or elsewhere without significant original commentary or change YouTube; it is judged across the "entire channel" YouTube, and it applies "even if you have permission" from the owner, because it is separate from copyright YouTube.
When does an AI video need a label?
Only when it looks real. In November 2023 YouTube announced that creators would have to disclose realistic altered or synthetic content, and that those who consistently fail to could face "suspension from the YouTube Partner Program" YouTube. It launched a "new tool in Creator Studio" for the disclosure on March 18, 2024 YouTube. Using AI for an outline or "script" needs no disclosure YouTube, and neither does "Cloning one’s own voice" for voiceovers or dubs YouTube. YouTube says a disclosure alone will not limit a video's reach or its "eligibility to earn money" YouTube. It may label videos itself when they were made with its own tools, carry "C2PA metadata" or are detected as AI YouTube, and since May 27, 2026 its systems "automatically apply a label" when they detect significant photorealistic AI use a creator did not declare YouTube. The current Help page repeats that creators who keep failing to disclose risk "suspension from the YouTube Partner Program" YouTube. Video made with YouTube's own Shorts tools carries "SynthID" watermarks YouTube.
For people whose faces or voices get copied, YouTube's likeness detection tool went to an "open beta" for partner-program creators in September 2025 YouTube, and in May 2026 YouTube said it would open to all creators "18 and older" YouTube. In his 2026 letter, YouTube's CEO named managing "AI slop" as a priority, building on YouTube's spam and clickbait systems YouTube.
How many channels does YouTube remove for spam?
Millions every quarter, and spam is most of it. In April to June 2026 YouTube removed 2,732,395 channels for Community Guidelines violations Google Transparency Report, 2,310,181 of them for spam, deceptive practices and scams Google. Taking those channels down removed another 86,686,063 videos with them Google, on top of the 11,159,676 videos removed one by one Google. Only 16,754 of the individually removed videos were taken down for spam Google: spam is handled by closing whole channels, not by pulling single videos. The report does not break out AI-generated, faceless or inauthentic channels, and YouTube publishes no count of channels demonetized under the inauthentic content policy.
Quarter
Channels removed for spam, misleading or deceptive practices and scams
Source: Google Transparency Report, YouTube Community Guidelines enforcement, read through the report's public data files. From April 2025 the category is named "Spam, deceptive practices and scams". The July-September 2025 spike came in the quarter of the inauthentic content rename, but the report counts all spam terminations and does not say why that quarter was high.
How much does YouTube pay creators, and how many share it?
More than $100 billion in four years, split across about three million partner channels and the music and media companies. YouTube said in January 2026 that it had paid over $100 billion to creators, artists and media companies in the previous four years YouTube, up from over $70 billion in the three years to February 2024 YouTube. Those payouts include record labels, publishers and broadcasters, not only individual channels. The YouTube Partner Program had more than 3 million channels in February 2024 YouTube and still "over 3 million creators" in August 2026 YouTube. YouTube does not publish how many faceless channels are in it, or how many channels earn a given amount; it says only that over 50% of partner channels earning five figures or more in 2024 also earned from sources other than ads and Premium YouTube.
Alphabet's filings give the advertising pool those payouts come from. YouTube ad revenue was 15,149 million dollars in 2019 Alphabet 10-K, 29,243 million in 2022 Alphabet 10-K and 40,367 million in 2025 Alphabet 10-K. Counting subscriptions, Alphabet said YouTube's revenue passed $60 billion in 2025 Alphabet. In the third quarter of 2026, YouTube ads brought in 11,207 million dollars Alphabet 10-Q, and YouTube Music and Premium passed 150 million subscribers including trials Alphabet.
Year
YouTube ad revenue (USD millions, company-reported)
Source: Alphabet Inc. annual reports on Form 10-K, filed with the SEC.
How much video is uploaded and watched?
YouTube's press page says, as of October 2026, that over 20 million videos are uploaded on an average day YouTube. Shorts averaged over 70 billion daily views in early 2024 YouTube and over 200 billion by June 2025 YouTube; part of that jump is a counting change, because since March 31, 2025 YouTube counts a Shorts view each time one starts or replays (YouTube Help). Viewers watch over 1B hours of YouTube a day on TV sets YouTube, and the number of top creators who get most of their watch time on TV rose more than 400% in the three years to early 2024 YouTube. That matters to faceless channels, most of whose videos in our sample are long ones.
Output is concentrated. In Pew Research Center's count of the 43,770 channels with at least 250,000 subscribers in late 2018 Pew, 10% of the channels produced 70% of the videos posted in the first week of 2019 Pew Research Center, 48,486 hours of video in seven days Pew. Our faceless sample shows the same skew: one channel posted nearly a fifth of all uploads.
Who watches YouTube?
84% of U.S. adults say they use YouTube, in Pew's 2026 survey Pew Research Center, up from 73% in early 2019 Pew, and 95% of adults aged 18 to 29 do Pew. Roughly nine-in-ten U.S. teens use it, and about one-in-five say they are on it almost constantly Pew, 2025, Pew. In the UK, 94% of adults use YouTube Ofcom and spend 51 minutes a day on it outside the TV set Ofcom.
Virtual creators: the faceless format YouTube does count
VTubers and other animated-avatar creators are the one faceless format YouTube reports on. Videos related to VTubers have averaged 50 billion views a year over the three years to 2024 YouTube Culture & Trends, and a sample of 300 virtual creators drew over 15 billion views in 2024 YouTube. In a YouTube-commissioned survey, 57% of U.S. online 14- to 44-year-olds said they had watched a virtual creator in the past year YouTube Culture & Trends; the report's footnotes give two possible sample sizes for this question. In YouTube's April 2026 survey, 72% of U.S. video posters aged 14 to 44 said they use AI on videos or images YouTube Culture & Trends, and 61% were interested in using AI to put themselves into content YouTube.
The economic footprint YouTube commissions
YouTube pays Oxford Economics to estimate its economic impact, so these are commissioned figures. Its 2025 U.S. estimate is over $60 billion in GDP Oxford Economics for YouTube and more than 540,000 full-time-equivalent jobs Oxford Economics for YouTube, against $55 billion in 2024 Oxford Economics. The job counts do not move in a straight line: more than 425,000 for 2021 Oxford Economics and more than 390,000 for 2022, on GDP of over $35 billionOxford Economics. In the UK, the 2025 report estimates £2.8 billion of GDP Oxford Economics for YouTube and more than 60,000 jobs Oxford Economics.
How many creators use AI to make their videos?
Most of them, by every survey that asked, though nobody has measured how many hide their face. YouTube's CEO said more than 1M channels on average used YouTube's own AI creation tools every day in December 2025 YouTube. In a 2024 survey YouTube commissioned from Radius, 92% of creators said they already used generative AI tools YouTube/Radius; YouTube did not publish the sample size. Adobe's survey of more than 16,000 creators in eight countries, run with The Harris Poll, found 86 percent using creative generative AI in September 2025 Adobe, most often for editing, upscaling and enhancement, then for generating new images and video (52 percent) Adobe. 60 percent had used more than one AI tool in three months Adobe. Adobe sells AI creative tools, so these are a vendor's figures.
Two years earlier the picture was less settled. In Descript's survey of 1,004 U.S. podcasters and video creators, run by Ipsos from March 29 to April 12, 2023 Descript, 78% said they were likely to use generative AI tools Descript, and 45% of those who had not tried them said they were not familiar with them Descript. Descript sells AI editing software.
Does AI help or hurt?
Creators who use it say it helps, and also that it makes it harder to be seen. In Adobe's May 2026 survey, 87% of creators who use or have tried creative AI said it had sped up the growth of their business or audience Adobe, against 76 percent of all creators surveyed in September 2025 Adobe; the two bases differ, so the rise is not like for like. 75% called it integrated or essential to how they work Adobe, and 40% said AI-assisted content consistently performs better Adobe. Of creators who find it harder to stand out than a year earlier, 42% blamed AI-generated content Adobe. On disclosure, 49% said they always or often tell their audience they used AI, and 18% rarely or never do Adobe.
AI voices and dubbing
Narration is the part of a faceless video AI most often replaces, and YouTube now builds it in. Its auto dubbing tool covered 27 languages by May 2026 YouTube, with Expressive Speech in 8 of them YouTube, and YouTube says "millions" of channels use it YouTube. In December 2025 more than 6 million viewers a day watched at least 10 minutes of auto-dubbed content YouTube. YouTube said in February 2025 that for videos with dubbed audio, over 40% of watch time came from viewers who chose a dubbed language YouTube, and creators who uploaded multi-language audio saw over 25% of watch time come from views in another language YouTube. YouTube does not require a disclosure for "Cloning one’s own voice" YouTube, or for an AI-written outline or "script", but it does for "AI generated music" and realistic AI footage YouTube.
The voice market is paying out. ElevenLabs, which sells AI voices, says creators on its Voice Library had earned over $22 million by May 2026 ElevenLabs, up from $1M in October 2024 ElevenLabs, with more than 10,400 voice creators earning ElevenLabs. It reported over $330 million in annual recurring revenue at the end of 2025 ElevenLabs. On the union side, SAG-AFTRA's January 2024 agreement with Replica Studios set a minimum of $956.75 for a four-hour session to create a digital voice replica SAG-AFTRA, archived, as a 1 year experiment SAG-AFTRA, capped training use at 3 years without new consent SAG-AFTRA, and requires replicas to be deleted when a contract or project ends; the word in the terms is "delete" SAG-AFTRA.
Stock footage, the other raw material
Faceless channels that do not generate their visuals license them. Shutterstock's library held 59 million footage clips at the end of 2024 Shutterstock 10-K, up from 54 million a year earlier Shutterstock 10-K, and over 3.5 million customers licensed its content in 2025 Shutterstock 10-K. Its data and distribution line, which includes licensing material for training AI models, earned $203.3 million in 2025 (reported as 203,264 thousand dollars) Shutterstock 10-K, but in the first half of 2026 its main content revenue fell to $343.8 million (reported as 343,790 thousand) from $402.7 million a year earlier Shutterstock 10-Q. The planned Getty Images merger was terminated on July 7, 2026Shutterstock. Getty Images owned or licensed over 645 million images and videos at the end of 2025 Getty Images 10-K.
How much of YouTube is AI-generated?
Nobody has counted it at platform scale; every figure is a sample. The most cited is Kapwing's October 2025 test, in which 104 of the first 500 Shorts shown to a brand-new YouTube account were AI-generated, and 165 were "brainrot" Kapwing. A similar Kapwing test on TikTok, published in June 2026, found 294 of the first 500 For You videos were AI slop Kapwing. Kapwing sells AI video tools, and one fresh account's feed is one feed. Kapwing also checked the top 100 trending channels in each country by hand: Pakistan had 20 AI slop channels in its top 100, Egypt 14, South Korea 11 and the U.S. nine Kapwing. The Guardian reported Kapwing's total as 278 AI-slop-only channels among about 15,000 top channels The Guardian, with more than 63bn views between them The Guardian and an estimated $117m a year in revenue The Guardian; those three totals do not appear on Kapwing's own report page.
The Guardian's own count of Playboard data found nine of the 100 fastest-growing YouTube channels in July 2025 showing purely AI-generated content The Guardian. After the paper contacted YouTube about these and other AI channels, some from June's fastest-growing list, YouTube said it had removed three and blocked two more from earning ad income The Guardian. NeoMam, the agency that produced Kapwing's study, re-checked the list after YouTube's CEO promised action on AI slop in January 2026 and found 16 of the top 100 most-subscribed slop channels gone NeoMam; it is reporting on its own campaign.
AI channels are new; the guides' examples are not
A 2026 comparison at Brazil's Federal University of Minas Gerais (an undergraduate thesis, not peer reviewed) set 100 explicitly AI-generated YouTube channels against 100 others. 63 of the AI channels were created in 2024 to 2026, against 3 of the comparison channels, and 59.5% of the AI channels' videos were Shorts, against 26.8% UFMG. That is the opposite of the channels the faceless guides hold up: in our read, the median example dates from 2016 and posts almost no Shorts. A separate study presented at the Association of Internet Researchers coded 194 of 382 filtered English-language videos about AI and children from 2022 to 2024 as AI-generated children's content AoIR Selected Papers, out of 9,277 English videos its first queries returned AoIR.
What do viewers think of AI-made video?
They want to know, and they doubt they can tell. 76% of U.S. adults told Pew Research Center in 2025 that it is extremely or very important to be able to tell whether pictures, videos and text were made by AI Pew Research Center, and 53% were not confident they could Pew. By 62% to 23%, Americans disapproved of online videos made with AI tools in YouGov's October 2025 poll YouGov. In the UK, 85% of adults support platforms labelling AI content, but only 34% have ever seen a label Ofcom, archived, and fewer than one in ten, 9%, are confident they can spot a deepfake Ofcom. 43% of UK people aged 16 and over said they had seen at least one deepfake in six months Ofcom. In a Common Sense Media survey fielded March to May 2024, 73% of U.S. teens wanted AI content labelled or watermarked Common Sense Media and 35% said they have been misled by fake content online Common Sense Media.
For news, comfort is low. Across six countries in 2025, only 19% were comfortable with AI being used to create an artificial presenter or author Reuters Institute and 12% with news made entirely by AI Reuters Institute. Weekly use of generative AI to make video was basically flat, at 3%Reuters Institute.
Detection is hard even in a lab. In a January 2026 test run by Runway, which makes AI video models, only 9.5% of 1,043 participants could reliably tell its five-second clips from real footage Runway. In a 2025-2026 UC Berkeley experiment with 400 U.S. participants, accuracy at spotting AI-generated clips rose from 58% to 75% as they ran from a single frame to 8 seconds, while accuracy on real clips stayed near 60% UC Berkeley.
How big are VTubers, the oldest faceless format?
Big in Japan, and unequal. COVER, which runs hololive, reported net sales of ¥49,330 million for the year to March 2026 COVER Corporation, with 95.02 million subscribers across its talents' channels COVER, 43 of its 84 VTubers past a million subscribers COVER, and 27.5% of viewers outside Japan COVER. ANYCOLOR, which runs NIJISANJI, reported ¥55,681 million for the year to April 2026 ANYCOLOR. Yano Research forecasts Japan's VTuber market at 1,260億円 (¥126.0 billion) in fiscal 2025 Yano Research Institute, up from 800億円 (¥80.0 billion) in fiscal 2023 Yano Research. ANYCOLOR's own survey puts the share of Japanese 15- to 39-year-olds who have watched VTuber streams at 24%ANYCOLOR.
Most VTubers earn little. A study of 1,923 VTubers' YouTube livestreams from 2017 to July 2023 arXiv preprint found average monthly livestream income of $2,667 but a median of $127 arXiv, with hololive and NIJISANJI taking more than 60% of all views arXiv and a Gini index of earnings that rose to 0.75 by 2023 arXiv.
What the popular faceless YouTube numbers really say
Course sellers, AI video tools and creator blogs repeat a small set of figures about faceless channels. We traced each to the earliest version we could find. For each we ran exact-phrase searches of the figure and its older wordings, checked Internet Archive captures of the pages that first carried it, read the trade press of the time and followed every "according to" back to its source. None of the percentages has a measured original, and the "AI ban" was a rename.
"38% of new creators are faceless" and "40% of the top 1,000 channels"
No one measured either. The earliest copy of the 38% we found is a vidBoard.ai blog post of April 8, 2025, which says 38% of new creator "monetization ventures" are faceless, citing only recent data vidBoard.ai. It counts YouTube and TikTok ventures, not channels. A "12% in 2022" baseline was added by AutoFaceless on December 21, 2025, credited to a vidBoard report that no vidBoard page contains AutoFaceless. From there it moved to Miraflow, still as 38%Miraflow AI, to Shortopus, which drew a series from 12% with 2023 and 2024 points no upstream page carries Shortopus, and to Frameloop, which turned the 38% into a share of new YouTube Partner Program entrants Frameloop. A different figure, 38.4% of channels with over 100,000 subscribers, is credited to Tubular Labs by a voice-tool blog Voxbooster; Tubular has no such page in its site or its archive.
The 40% figure for YouTube's top 1,000 channels first appears in an OutlierKit guide published in February 2026, with no method OutlierKit. OutlierKit has since dropped it: its guide now says the share is "genuinely hard to pin down" because faceless has no standard definition, for the "top 1,000" or any other list OutlierKit. Another site still credited OutlierKit with "around 40%" on October 14, 2026 Faceless Channel. What to use instead: no platform-wide count of faceless channels exists. YouTube does not classify channels that way, and the closest real counts are small samples of AI channels (see above).
"Faceless channels make $10,000 a month"
A sales target, not a measurement. The most concrete source is a course seller's own testimonials: Faceless Channel Academy lists student results "as reported to us", led by $10,000 a month, and calls them not typical Become Viral. NexLev, which sells YouTube automation software, tells readers in its 2026 guide that they "can realistically make $5,000 to $10,000 a month" NexLev; Frameloop stretches it to "$5,000 to $50,000" Frameloop. vidBoard.ai says faceless channels "frequently earn $15-40 per 1,000 views" with no source vidBoard.ai, while its own monetization guide gives $3-12 RPM vidBoard.ai. The dollar figures for named channels in most listicles come from Social Blade, which multiplies views by a fixed rate range whose low end is $0.25 per 1,000 Social Blade.
What to use instead: YouTube defines RPM as what one channel earned per 1,000 views YouTube Help, after its revenue share and counting views that earned nothing, which is why "RPM" sits below CPM YouTube Help. Long-form partners keep 55% of ad revenue YouTube. It varies by channel and niche; our YouTube CPM and RPM page collects what creators have disclosed.
"YouTube is demonetizing AI channels" (July 2025)
It renamed a policy. YouTube's policy page records the change of July 15, 2025 as "a minor update" renaming "repetitious content" to "inauthentic content" YouTube Help, and says the content "has always been ineligible" for monetization YouTube Help. YouTube Liaison Rene Ritchie called it a "minor update" on July 8 TechCrunch and said such content had been ineligible "for years" TechCrunch. The July 17, 2025 capture of the policy page, two days after the change, carries the July 15, 2025 note and does not mention AI at all YouTube Help, archived. Headlines said otherwise: Black Enterprise reported YouTube demonetizing channels that rely entirely on "AI-generated content" Black Enterprise, EWN said it would stop monetising AI videos from "15 July 2025" EWN, and The Star in Kenya headlined the end of monetization for "AI-generated videos" The Star.
AI entered the policy only in mid-2026, in two narrow cases: content made with generic AI templates, and AI personas posing as a "human expert" on health, legal, financial or political topics YouTube Help. YouTube's VP of Trust and Safety said that July 2026 breakdown announced no new policy for "YPP" channels Net Influencer. What to use instead: the policy wording dated above, and the fact that YouTube publishes no count of channels demonetized under it.
"YouTube terminated 16 AI channels with 4.7 billion views"
The count is Kapwing's, not YouTube's. A June 2026 post says YouTube terminated 16 channels with 4.7 billion views in April 2026 in what the platform "internally called" an AI slop enforcement wave Spokio. The number comes from Kapwing re-checking its own top-100 slop list in January 2026, which found 16 channels gone Net Influencer with 4,722,138,437 views between them, based on Social Blade Net Influencer. YouTube published no such count and no YouTube source uses the word "internally".
"Shorts pay $0.03 per 1,000 views"
One channel's Short. YouTube publishes no Shorts RPM; it says partners keep 45% of the Shorts revenue allocated to them YouTube Help. vidIQ's range of $0.03 to $0.10 rests on one Short from its own channel vidIQ. AIR Media-Tech's data from 274 of its partner channels puts Shorts RPM between $0.02 and $1.48 depending on niche AIR Media-Tech, a range wide enough that no single figure describes Shorts pay.
We started from the questions people search for about faceless youtube statistics and looked for the organization that actually produced each answer. Every figure on this page sits in an evidence ledger with its source, the exact table or paragraph it came from, the measurement period, the population measured, and the date we checked it. This edition holds 239 figures from 42 publishers.
On October 31, 2026, each of the sourced figures on this page was re-fetched with the fetch cache cleared and matched against the recorded wording: YouTube's Help Center, Official Blog and How YouTube Works pages, dated Internet Archive captures of YouTube's monetization policy pages for the wording changes, the Google Transparency Report's public data files, Alphabet, Shutterstock and Getty Images filings on SEC EDGAR, survey owners' own reports (Pew Research Center, Reuters Institute, Ofcom, YouGov, Common Sense Media, Adobe, Descript) and the papers and company reports cited. Our own figures come from a YouTube Data API read on October 31, 2026 of the 126 channels that eight public faceless-channel guides link as examples, covering uploads from October 31, 2025 to October 30, 2026; the per-channel and per-video data is published in channels.csv and videos.csv. We left out Kapwing's channel totals as primary figures (they appear only in The Guardian's report, not on Kapwing's page), a New York Times analysis behind a paywall, a ReelForge claim whose text loads only with JavaScript, and YouTube Help pages we could not read without a browser.
We only cite the producer: government agencies, regulators, company filings and official posts, and the survey owner's own report. We do not cite statistics roundups, Statista-style aggregators, or salary sites without a public method. We did not bypass paywalls or sign-up forms; where a full report was gated, we used only the figures its publisher states publicly.
Figures marked "our calculation" are arithmetic on the cited numbers, and anyone can reproduce them:
This edition contains no Mark Studios customer data.
Limitations
Our sample is the guides' choice, not a random draw. The 126 channels are the ones eight creator-tool and marketing blogs link as faceless examples. A few show a presenter, AI-only channels are barely represented, and several of the publishers sell tools for making faceless video. The read says what those examples look like, not what all faceless channels do.
No one counts faceless or AI channels across YouTube. YouTube does not classify channels as faceless, does not publish how many are demonetized under the inauthentic content policy, and its public Data API documents the AI disclosure only as a field channel owners set when uploading or editing; it returned that field for none of the 15,509 uploads we read. Kapwing's, The Guardian's and the academic counts are samples with their own definitions.
Policy pages change without notice. Two 2026 changes to YouTube's monetization policy carried no dated note; we dated them from Internet Archive captures, so the true change falls somewhere between two captures.
Many figures come from parties with a stake. YouTube reports on its own platform and commissions the Oxford Economics, Radius and Culture & Trends studies; Adobe, Descript, Kapwing, ElevenLabs and Runway sell AI or video tools; COVER and ANYCOLOR report on their own business. Each is labelled where it appears.
Several figures are snapshots. Subscriber and view counts, enforcement totals and policy wording are as read on the dates given.
How to cite this page
For a single figure, cite the original publisher named next to it, and link here if the verification trail is useful to your readers. Each figure has its own anchor, so you can link straight to it, for example https://www.markstudios.com/research/faceless-youtube-statistics#DRV-008. To cite the page as a whole:
Mark Studios. "Faceless YouTube Statistics." Last verified Oct 31, 2026. https://www.markstudios.com/research/faceless-youtube-statistics
APA
Mark Studios. (2026, October 31). Faceless YouTube statistics. https://www.markstudios.com/research/faceless-youtube-statistics
MLA
“Faceless YouTube Statistics.” Mark Studios, 31 Oct. 2026, www.markstudios.com/research/faceless-youtube-statistics.
Chicago
Mark Studios. “Faceless YouTube Statistics.” Last modified October 31, 2026. https://www.markstudios.com/research/faceless-youtube-statistics.
Our calculations and the evidence table are free to reuse with attribution and a link to this page. Each underlying figure remains the claim of its original publisher, and some publishers ask for a link when their figures are reused.
We re-verify every figure quarterly and when a major source publishes a new edition. The next scheduled review is January 2027. Corrections go to contact@markstudios.com.
Oct 31, 2026: First published with 239 figures from 42 publishers, each checked against its original source.
Oct 31, 2026: Corrected several descriptions after a source-by-source check. The fan-funding tier also needs 3,000 watch hours or 3 million Shorts views. Existing partners accept the 2027 terms to keep fully monetizing. Kapwing does not say Pakistan had the most AI slop channels of any country. YouTube's three removals after The Guardian's inquiry included channels beyond July's list. UC Berkeley's 58% to 75% is for spotting AI clips; accuracy on real clips stayed near 60%. SAG-AFTRA's $956.75 is a session minimum. Reuters' 3% for video was basically flat, and its 19% covers artificial presenters or authors. Faceless Channel's wording is "around 40%". The Data API lists the AI disclosure only as a field owners set, and the 40% dubbed-audio share is from YouTube's February 2025 letter.
Oct 31, 2026: Removed an unsourced aside about when AI voice and video tools became cheap, and cited YouTube's help page for the March 31, 2025 change to how Shorts views are counted.
Frequently asked questions
Can faceless YouTube channels be monetized?
Yes. YouTube's rules do not mention faces. As of October 2026 a channel needs 1,000 subscribers plus 4,000 watch hours in 12 months, or 10 million Shorts views in 90 days, and its content must not be mass-produced, repetitive or reused without significant change.
Did YouTube ban AI or faceless channels in July 2025?
No. On July 15, 2025 YouTube renamed its repetitious content policy to inauthentic content and said such content had always been ineligible for monetization. The policy page did not mention AI until 2026.
What are the new YouTube Partner Program requirements for 2027?
From February 1, 2027, new applicants need 8,000 qualified watch hours in the last 365 days or 20M qualified Shorts views in 90 days, plus 1,000 subscribers. Channels already in the program are not affected.
Do faceless YouTube channels make $10,000 a month?
There is no measured basis for that figure. The most concrete source is a course seller's own student testimonials, led by $10,000 a month and described as not typical. YouTube publishes no earnings figure for faceless channels.
What percentage of YouTube channels are faceless?
Nobody knows. YouTube does not classify channels as faceless. The widely repeated 38% traces to a 2025 vendor blog post with no data, and the 40% of top channels figure was dropped by the site that first published it.
How much of YouTube is AI-generated?
No platform-wide count exists. In Kapwing's October 2025 test, 104 of the first 500 Shorts shown to a new account were AI-generated; a Guardian analysis found nine of the 100 fastest-growing channels in July 2025 showed purely AI-generated content.
Do I have to label AI-generated videos on YouTube?
Only realistic altered or synthetic content. YouTube does not require a label for AI-written scripts or for cloning your own voice, and since May 27, 2026 it can automatically apply a label when it detects significant photorealistic AI use.
