YouTube's Community Guidelines explain what content isn't allowed on the platform and how violations are enforced — covering spam, sensitive content, violent content, regulated goods, and misinformation, plus the strike system for repeat violations. The two keywords tracked here (machine learning, train AI/models) both point to the same enforcement-pipeline disclosure: YouTube uses machine-learning systems, alongside human reviewers, to detect and remove policy-violating content at scale, and explicitly states that human reviewers' moderation decisions are fed back in to train and improve those same ML systems ("Reviewers' inputs are then used to train and improve the accuracy of our systems at a much larger scale"). This is a content-moderation AI disclosure, not a user-content-for-generative-AI clause — YouTube is not saying it trains a generative model on uploaded videos here, only that it trains its own spam/abuse-detection classifiers on its moderators' decisions.
What this wiki found — complete, every page
Presence facts, not verdicts — each flagged term links to its definition and the exact clause on this page. Absence of a badge means the term isn't currently flagged here, not that the page is risk-free.
Independent reviews, where they had something to say
YouTube – Community Guidelines
Dataset: PGAv2 (Platform Governance Archive v2) — 25 platforms, 2022–2026
PGAv2 (Platform Governance Archive v2) is this wiki’s legacy dataset: pre-tagged JSONL records covering 25 major social-media, messaging, and content-sharing platforms, with risk scores and keywords assigned via systematic extraction of 1,736 high-risk clauses from each platform’s Terms of Service, Privacy Policies, and Community Guidelines, spanning 2022–2026.
Note: This page contains documented policy clauses. Risk assessment is qualitative and context-dependent. For analysis of patterns across platforms, see:
Source: sources/jsonl/YouTube_Community_Guidelines.jsonl
Last updated: 2025-10-29
Overview
YouTube’s Community Guidelines explain what content isn’t allowed on the platform and how violations are enforced — covering spam, sensitive content, violent content, regulated goods, and misinformation, plus the strike system for repeat violations. The two keywords tracked here (machine learning, train AI/models) both point to the same enforcement-pipeline disclosure: YouTube uses machine-learning systems, alongside human reviewers, to detect and remove policy-violating content at scale, and explicitly states that human reviewers’ moderation decisions are fed back in to train and improve those same ML systems (“Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale”). This is a content-moderation AI disclosure, not a user-content-for-generative-AI clause — YouTube is not saying it trains a generative model on uploaded videos here, only that it trains its own spam/abuse-detection classifiers on its moderators’ decisions.
This page covers an unusually long and stable scrape history: 63 scrape dates from 2022-11-02 to 2025-10-29, and the underlying AI/ML clause text has been word-for-word identical in every single scrape. Across the whole 3-year window, only 4 dates introduce any new captured text at all (2023-08-24, 2023-09-12, 2025-02-12, 2025-02-14), and none of those edits touch the actual machine-learning/training sentences — they are a misinformation-policy link consolidation, a new cross-reference link addition, and two markdown heading-level rendering artifacts (see Changes Summary). In total there are only 13 unique snippets in the entire 390-record file.
Flagged Keywords & Risks
machine learning— Tags multiple passages describing YouTube’s moderation pipeline: “we combine the power of advanced machine learning systems and our community itself to flag potentially problematic content,” and “Machine learning systems help us identify and remove spam automatically.” Why it matters: this discloses that automated systems — not just human judgment — make or assist first-pass decisions about what content gets flagged or removed, with limited visibility into the system’s accuracy or appeal process for creators caught by it.train AI/models— From the same passage: “Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale.” Why it matters: this is a direct admission that human moderators’ decisions become training data for YouTube’s automated enforcement models — a legitimate and fairly common feedback-loop design for trust & safety systems, but one that means a moderator’s individual judgment calls can get encoded into automated decisions affecting many other creators later.
Legal Context & Research Significance
The “Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale” sentence is a content-moderation training disclosure rather than a generative-AI-on-user-content clause, and it falls under wearetosed’s “tracking & profiling” category insofar as YouTube’s machine-learning systems make first-pass decisions about content with limited visibility into accuracy or appeal for the creators affected. Javed & Sajid (2024) found that only 42.57% of the 202 privacy-policy-analysis papers they reviewed used machine learning, deep learning, or NLP to automate extraction of clauses like this one — the remainder relied on manual or qualitative methods — which is notable given that this page’s own clause has stayed byte-for-byte identical across all 63 scrapes from 2022-11-02 to 2025-10-29, the kind of long stable run that automated extraction tools are well suited to verify at scale. Atkinson’s (2025) contract-notice framework for scraping/training prohibitions does not bear directly on this page, since the clause describes YouTube training its own moderation classifiers internally rather than restricting outside parties’ use of YouTube content.
Changes Summary
| Date | What changed |
|---|---|
| 2022-11-02 | Baseline version (first scrape in this dataset) — 6 unique snippets, both keywords present. |
| 2022-12-01 through 2023-08-15 | No changes from previous version (17 consecutive scrapes). |
| 2023-08-24 | machine learning updated wording — but the only difference is in the surrounding bullet list, not the AI clause: the separate “COVID-19 medical misinformation” and “Vaccine misinformation” help-center links were consolidated into a single “Medical misinformation” link. |
| 2023-08-29 | No changes from previous version. |
| 2023-09-12 | machine learning new snippet added — a new “Educational, Documentary, Scientific, and Artistic (EDSA) content” cross-reference link was added to both the policy-overview bullet list and the “Resources” section. No AI/ML clause text changed. |
| 2023-09-27 through 2025-01-31 | No changes from previous version (~30 consecutive scrapes). |
| 2025-02-12 | machine learning updated wording — pure markdown heading-level rendering difference (### → # on a bullet-list heading); no text content changed. |
| 2025-02-14 | machine learning updated wording — markdown heading level reverted again (#### → no heading marker); no text content changed. |
| 2025-02-19 through 2025-10-29 | No changes from previous version (covers the remaining 16 scrapes through 2025-10-29). |
Version History
2022-11-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
Clause: machine learning
Community Guidelines ==================== Overview -------- YouTube has always had a set of Community Guidelines that outline what type of content isn’t allowed on YouTube. These policies apply to all types of content on our platform, including videos, comments, links, and thumbnails. Our Community Guidelines are a key part of our broader suite of policies and are regularly updated in consultation with outside experts and YouTube creators to keep pace with emerging challenges. We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally—regardless of the subject or the creator’s background, political viewpoint, position, or affiliation. Our policies aim to make YouTube a safer community while still giving creators the freedom to share a broad range of experiences and perspectives. What areas do Community Guidelines cover? ----------------------------------------- You’ll find a full list of our Community Guidelines below: ### Spam & deceptive practices * Fake engagement * Impersonation * External links * Spam, deceptive practices & scams * Playlists * Additional policies ### Sensitive content * Child safety * Thumbnails * Nudity and sexual content * Suicide and self-harm * Vulgar language ### Violent or dangerous content * Harassment and cyberbullying * Harmful or dangerous content * Hate speech * Violent criminal organizations * Violent or graphic content ### Regulated goods * Firearms * Sale of illegal or regulated goods or services ### Misinformation * Misinformation * Elections misinformation * COVID-19 medical misinformation * Vaccine misinformation In addition to Community Guidelines, creators who want to monetize content on YouTube must comply with Monetization Policies. How does YouTube develop new policies and update existing ones?
Clause: machine learning
This work is never finished, and we are always evaluating our policies to understand how we can better strike a balance between keeping the YouTube community protected and giving everyone a voice. How does YouTube enforce its Community Guidelines? -------------------------------------------------- 500 hours of video are uploaded to YouTube every minute. That’s a lot of content, which is why our teams come together to make sure that what you see on our platform follows our Community Guidelines. To do that, we combine the power of advanced machine learning systems and our community itself to flag potentially problematic content. Our expert reviewers then remove flagged content that violates our Community Guidelines. How does YouTube identify content that violates Community Guidelines? --------------------------------------------------------------------- With hundreds of hours of new content uploaded to YouTube every minute, we use a combination of people and machine learning to detect problematic content at scale. Machine learning is well-suited to detect patterns, which helps us to find content similar to other content we’ve already removed, even before it’s viewed.
Clause: machine learning
That’s a lot of content, which is why our teams come together to make sure that what you see on our platform follows our Community Guidelines. To do that, we combine the power of advanced machine learning systems and our community itself to flag potentially problematic content. Our expert reviewers then remove flagged content that violates our Community Guidelines. How does YouTube identify content that violates Community Guidelines? --------------------------------------------------------------------- With hundreds of hours of new content uploaded to YouTube every minute, we use a combination of people and machine learning to detect problematic content at scale. Machine learning is well-suited to detect patterns, which helps us to find content similar to other content we’ve already removed, even before it’s viewed. We also recognize that the best way to quickly remove content is to anticipate problems before they emerge. Our Intelligence Desk monitors the news, social media, and user reports to detect new trends surrounding inappropriate content, and works to make sure our teams are prepared to address them before they can become a larger issue. Is there a way for the broader community to flag harmful content?
Clause: machine learning
To do that, we combine the power of advanced machine learning systems and our community itself to flag potentially problematic content. Our expert reviewers then remove flagged content that violates our Community Guidelines. How does YouTube identify content that violates Community Guidelines? --------------------------------------------------------------------- With hundreds of hours of new content uploaded to YouTube every minute, we use a combination of people and machine learning to detect problematic content at scale. Machine learning is well-suited to detect patterns, which helps us to find content similar to other content we’ve already removed, even before it’s viewed. We also recognize that the best way to quickly remove content is to anticipate problems before they emerge. Our Intelligence Desk monitors the news, social media, and user reports to detect new trends surrounding inappropriate content, and works to make sure our teams are prepared to address them before they can become a larger issue. Is there a way for the broader community to flag harmful content? ----------------------------------------------------------------- The YouTube community also plays an important role in flagging content they think is inappropriate.
Clause: machine learning
We often refer to this exception as “EDSA,” which stands for “Educational, Documentary, Scientific or Artistic”. To help determine whether a video might qualify for an EDSA exception, we look at multiple factors, including the video title, descriptions, and the context provided. EDSA exceptions are a critical way we make sure that important speech stays on YouTube, while protecting the wider YouTube ecosystem from harmful content. ### Resources * Read more about how we treat EDSA content on YouTube What action does YouTube take for content that violates Community Guidelines? ----------------------------------------------------------------------------- Machine learning systems help us identify and remove spam automatically, as well as remove re-uploads of content we’ve already reviewed and determined violates our policies. YouTube takes action on other flagged videos after review by trained human reviewers. They assess whether the content does indeed violate our policies, and protect content that has an educational, documentary, scientific, or artistic purpose. Our reviewer teams remove content that violates our policies and age-restrict content that may not be appropriate for all audiences. Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale.
Clause: train AI/models
----------------------------------------------------------------------------- Machine learning systems help us identify and remove spam automatically, as well as remove re-uploads of content we’ve already reviewed and determined violates our policies. YouTube takes action on other flagged videos after review by trained human reviewers. They assess whether the content does indeed violate our policies, and protect content that has an educational, documentary, scientific, or artistic purpose. Our reviewer teams remove content that violates our policies and age-restrict content that may not be appropriate for all audiences. Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale. Community Guidelines Strikes ---------------------------- If our reviewers decide that content violates our Community Guidelines, we remove the content and send a notice to the Creator. The first time a Creator violates our Community Guidelines, they receive a warning with no penalty to the channel. After one warning, we’ll issue a Community Guidelines strike to the channel and the account will have temporary restrictions including not being allowed to upload videos, live streams, or stories for a 1-week period. Channels that receive three strikes within a 90-day period will be terminated.
2022-12-01
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2022-11-02 (see above).
2022-12-06
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2022-12-01 (see above).
2022-12-09
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2022-12-06 (see above).
2023-03-25
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2022-12-09 (see above).
2023-03-28
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-03-25 (see above).
2023-04-15
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-03-28 (see above).
2023-04-18
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-04-15 (see above).
2023-04-19
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-04-18 (see above).
2023-04-20
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-04-19 (see above).
2023-06-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-04-20 (see above).
2023-06-28
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-06-02 (see above).
2023-06-30
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-06-28 (see above).
2023-07-13
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-06-30 (see above).
2023-07-20
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-07-13 (see above).
2023-08-10
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-07-20 (see above).
2023-08-15
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-08-10 (see above).
2023-08-24
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
train AI/models unchanged from 2023-08-15.
Clause: machine learning (updated wording — see Changes Summary; misinformation-policy link consolidation, not an AI/ML clause change)
Community Guidelines ==================== Overview -------- YouTube has always had a set of Community Guidelines that outline what type of content isn’t allowed on YouTube. These policies apply to all types of content on our platform, including videos, comments, links, and thumbnails. Our Community Guidelines are a key part of our broader suite of policies and are regularly updated in consultation with outside experts and YouTube creators to keep pace with emerging challenges. We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally—regardless of the subject or the creator’s background, political viewpoint, position, or affiliation. Our policies aim to make YouTube a safer community while still giving creators the freedom to share a broad range of experiences and perspectives. What areas do Community Guidelines cover? ----------------------------------------- You’ll find a full list of our Community Guidelines below: ### Spam & deceptive practices * Fake engagement * Impersonation * External links * Spam, deceptive practices & scams * Playlists * Additional policies ### Sensitive content * Child safety * Thumbnails * Nudity and sexual content * Suicide and self-harm * Vulgar language ### Violent or dangerous content * Harassment and cyberbullying * Harmful or dangerous content * Hate speech * Violent criminal organizations * Violent or graphic content ### Regulated goods * Firearms * Sale of illegal or regulated goods or services ### Misinformation * Misinformation * Elections misinformation * Medical misinformation In addition to Community Guidelines, creators who want to monetize content on YouTube must comply with Monetization Policies. How does YouTube develop new policies and update existing ones?
2023-08-29
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-08-24 (see above).
2023-09-12
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
train AI/models unchanged from 2023-08-29.
Clause: machine learning (new snippet added — see Changes Summary; new EDSA cross-reference link, not an AI/ML clause change)
Community Guidelines ==================== Overview -------- YouTube has always had a set of Community Guidelines that outline what type of content isn’t allowed on YouTube. These policies apply to all types of content on our platform, including videos, comments, links, and thumbnails. Our Community Guidelines are a key part of our broader suite of policies and are regularly updated in consultation with outside experts and YouTube creators to keep pace with emerging challenges. We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally—regardless of the subject or the creator’s background, political viewpoint, position, or affiliation. Our policies aim to make YouTube a safer community while still giving creators the freedom to share a broad range of experiences and perspectives. What areas do Community Guidelines cover? ----------------------------------------- You’ll find a full list of our Community Guidelines below: ### Spam & deceptive practices * Fake engagement * Impersonation * External links * Spam, deceptive practices & scams * Playlists * Additional policies ### Sensitive content * Child safety * Thumbnails * Nudity and sexual content * Suicide and self-harm * Vulgar language ### Violent or dangerous content * Harassment and cyberbullying * Harmful or dangerous content * Hate speech * Violent criminal organizations * Violent or graphic content ### Regulated goods * Firearms * Sale of illegal or regulated goods or services ### Misinformation * Misinformation * Elections misinformation * Medical misinformation ### Educational, Documentary, Scientific, and Artistic (EDSA) content * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content In addition to Community Guidelines, creators who want to monetize content on YouTube must comply with Monetization Policies. How does YouTube develop new policies and update existing ones?
Clause: machine learning (new snippet added — see Changes Summary; new EDSA cross-reference link, not an AI/ML clause change)
We often refer to this exception as “EDSA,” which stands for “Educational, Documentary, Scientific or Artistic”. To help determine whether a video might qualify for an EDSA exception, we look at multiple factors, including the video title, descriptions, and the context provided. EDSA exceptions are a critical way we make sure that important speech stays on YouTube, while protecting the wider YouTube ecosystem from harmful content. ### Resources * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content * Read more about how we treat EDSA content on YouTube What action does YouTube take for content that violates Community Guidelines? ----------------------------------------------------------------------------- Machine learning systems help us identify and remove spam automatically, as well as remove re-uploads of content we’ve already reviewed and determined violates our policies. YouTube takes action on other flagged videos after review by trained human reviewers. They assess whether the content does indeed violate our policies, and protect content that has an educational, documentary, scientific, or artistic purpose. Our reviewer teams remove content that violates our policies and age-restrict content that may not be appropriate for all audiences. Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale.
2023-09-27
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-09-12 (see above).
2023-09-28
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-09-27 (see above).
2023-09-30
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-09-28 (see above).
2023-10-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-09-30 (see above).
2023-11-07
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-10-02 (see above).
2023-12-04
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-11-07 (see above).
2023-12-12
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-12-04 (see above).
2023-12-14
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-12-12 (see above).
2023-12-19
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-12-14 (see above).
2024-02-22
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2023-12-19 (see above).
2024-03-08
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-02-22 (see above).
2024-05-01
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-03-08 (see above).
2024-05-06
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-05-01 (see above).
2024-06-13
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-05-06 (see above).
2024-06-21
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-06-13 (see above).
2024-08-22
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-06-21 (see above).
2024-11-06
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-08-22 (see above).
2024-11-20
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-11-06 (see above).
2024-12-11
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-11-20 (see above).
2024-12-21
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-12-11 (see above).
2025-01-15
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2024-12-21 (see above).
2025-01-29
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-01-15 (see above).
2025-01-30
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-01-29 (see above).
2025-01-31
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-01-30 (see above).
2025-02-01
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-01-31 (see above).
2025-02-12
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
train AI/models unchanged from 2025-02-01.
Clause: machine learning (updated wording — see Changes Summary; markdown heading-level rendering artifact, not a content change)
Community Guidelines ==================== Overview -------- YouTube has always had a set of Community Guidelines that outline what type of content isn’t allowed on YouTube. These policies apply to all types of content on our platform, including videos, comments, links, and thumbnails. Our Community Guidelines are a key part of our broader suite of policies and are regularly updated in consultation with outside experts and YouTube creators to keep pace with emerging challenges. We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally—regardless of the subject or the creator’s background, political viewpoint, position, or affiliation. Our policies aim to make YouTube a safer community while still giving creators the freedom to share a broad range of experiences and perspectives. What areas do Community Guidelines cover? ----------------------------------------- You’ll find a full list of our Community Guidelines below: #### Spam & deceptive practices * Fake engagement * Impersonation * External links * Spam, deceptive practices & scams * Playlists * Additional policies #### Sensitive content * Child safety * Thumbnails * Nudity and sexual content * Suicide and self-harm * Vulgar language #### Violent or dangerous content * Harassment and cyberbullying * Harmful or dangerous content * Hate speech * Violent criminal organizations * Violent or graphic content #### Regulated goods * Firearms * Sale of illegal or regulated goods or services #### Misinformation * Misinformation * Elections misinformation * Medical misinformation #### Educational, Documentary, Scientific, and Artistic (EDSA) content * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content In addition to Community Guidelines, creators who want to monetize content on YouTube must comply with Monetization Policies. How does YouTube develop new policies and update existing ones?
Clause: machine learning (updated wording — see Changes Summary; markdown heading-level rendering artifact, not a content change)
We often refer to this exception as “EDSA,” which stands for “Educational, Documentary, Scientific or Artistic”. To help determine whether a video might qualify for an EDSA exception, we look at multiple factors, including the video title, descriptions, and the context provided. EDSA exceptions are a critical way we make sure that important speech stays on YouTube, while protecting the wider YouTube ecosystem from harmful content. #### Resources * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content * Read more about how we treat EDSA content on YouTube What action does YouTube take for content that violates Community Guidelines? ----------------------------------------------------------------------------- Machine learning systems help us identify and remove spam automatically, as well as remove re-uploads of content we’ve already reviewed and determined violates our policies. YouTube takes action on other flagged videos after review by trained human reviewers. They assess whether the content does indeed violate our policies, and protect content that has an educational, documentary, scientific, or artistic purpose. Our reviewer teams remove content that violates our policies and age-restrict content that may not be appropriate for all audiences. Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale.
2025-02-14
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
train AI/models unchanged from 2025-02-12.
Clause: machine learning (updated wording — see Changes Summary; markdown heading-level rendering artifact, not a content change)
Community Guidelines ==================== Overview -------- YouTube has always had a set of Community Guidelines that outline what type of content isn’t allowed on YouTube. These policies apply to all types of content on our platform, including videos, comments, links, and thumbnails. Our Community Guidelines are a key part of our broader suite of policies and are regularly updated in consultation with outside experts and YouTube creators to keep pace with emerging challenges. We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally—regardless of the subject or the creator’s background, political viewpoint, position, or affiliation. Our policies aim to make YouTube a safer community while still giving creators the freedom to share a broad range of experiences and perspectives. What areas do Community Guidelines cover? ----------------------------------------- You’ll find a full list of our Community Guidelines below: Spam & deceptive practices * Fake engagement * Impersonation * External links * Spam, deceptive practices & scams * Playlists * Additional policies Sensitive content * Child safety * Thumbnails * Nudity and sexual content * Suicide and self-harm * Vulgar language Violent or dangerous content * Harassment and cyberbullying * Harmful or dangerous content * Hate speech * Violent criminal organizations * Violent or graphic content Regulated goods * Firearms * Sale of illegal or regulated goods or services Misinformation * Misinformation * Elections misinformation * Medical misinformation Educational, Documentary, Scientific, and Artistic (EDSA) content * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content In addition to Community Guidelines, creators who want to monetize content on YouTube must comply with Monetization Policies. How does YouTube develop new policies and update existing ones?
Clause: machine learning (updated wording — see Changes Summary; markdown heading-level rendering artifact, not a content change)
We often refer to this exception as “EDSA,” which stands for “Educational, Documentary, Scientific or Artistic”. To help determine whether a video might qualify for an EDSA exception, we look at multiple factors, including the video title, descriptions, and the context provided. EDSA exceptions are a critical way we make sure that important speech stays on YouTube, while protecting the wider YouTube ecosystem from harmful content. Resources * How YouTube evaluates Educational, Documentary, Scientific, and Artistic (EDSA) content * Read more about how we treat EDSA content on YouTube What action does YouTube take for content that violates Community Guidelines? ----------------------------------------------------------------------------- Machine learning systems help us identify and remove spam automatically, as well as remove re-uploads of content we’ve already reviewed and determined violates our policies. YouTube takes action on other flagged videos after review by trained human reviewers. They assess whether the content does indeed violate our policies, and protect content that has an educational, documentary, scientific, or artistic purpose. Our reviewer teams remove content that violates our policies and age-restrict content that may not be appropriate for all audiences. Reviewers’ inputs are then used to train and improve the accuracy of our systems at a much larger scale.
2025-02-19
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-02-14 (see above).
2025-03-05
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-02-19 (see above).
2025-03-29
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-03-05 (see above).
2025-04-01
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-03-29 (see above).
2025-04-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-04-01 (see above).
2025-04-17
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-04-02 (see above).
2025-06-25
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-04-17 (see above).
2025-07-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-06-25 (see above).
2025-07-26
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-07-02 (see above).
2025-08-20
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-07-26 (see above).
2025-08-28
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-08-20 (see above).
2025-09-23
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-08-28 (see above).
2025-10-02
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-09-23 (see above).
2025-10-03
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-10-02 (see above).
2025-10-23
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-10-03 (see above).
2025-10-29
- Explicit AI clause: YES
- Non-explicit predatory: NO
- Flagged keywords:
machine learning,train AI/models
All clauses unchanged from 2025-10-23 (see above).