# Labeling AI-Generated Content: China's Rules vs Platform Policies

> As of October 2026, China's Measures for Labeling AI-Generated and Synthetic Content require AI-made text, images, audio and video to carry labels: AI tools add a visible label where content could mislead and an implicit one to the file metadata, platforms check and display them, posters declare AI content, and no one may strip a label. Douyin, Xiaohongshu and WeChat Channels write the duty into their terms, while TikTok, YouTube and Meta require labels on realistic AI content and may add labels automatically when files carry C2PA or similar signals.

- URL: https://www.mvpdata.net/en/insights/ai-generated-content-labels
- Updated: 2026-10-17
- Published: 2026-10-17
- 简体中文: https://www.mvpdata.net/insights/ai-generated-content-labels

## Key takeaways

- China's labeling measures and the mandatory national standard GB 45438-2025 both took effect on September 1, 2025.
- An explicit label is a notice users can see or hear; an implicit label sits in the file metadata and records that the content is AI-generated, the provider and a content ID.
- Article 10 of the measures puts two duties on posters: declare AI content and label it with the platform tool, and never maliciously remove, alter, forge or hide a label.
- Livestream sellers in China who use AI-generated people must label them and keep shoppers informed throughout; refusing to correct a failure, or a serious case, can bring fines of 10,000 to 100,000 yuan under Order No. 117 where other laws and regulations are silent.
- TikTok and YouTube both state that labeling content as AI-made does not by itself reduce its distribution; YouTube adds that it does not affect monetization.

## What does China require for labeling AI-generated content?

Since September 1, 2025, AI-generated text, images, audio, video and virtual scenes created or shared in China must carry labels: AI tools add a visible label where content could mislead and an implicit label to the file metadata, platforms check and display them, and the people who post the content declare it.[^1] The rules are the Measures for Labeling AI-Generated and Synthetic Content (Guoxinban Tongzi [2025] No. 2), issued jointly on March 14, 2025 by the Cyberspace Administration of China (CAC), the Ministry of Industry and Information Technology, the Ministry of Public Security and the National Radio and Television Administration.[^2] Article 14 sets the effective date of September 1, 2025.[^3]

A mandatory national standard sits alongside them: GB 45438-2025, *Cybersecurity technology: Labeling method for content generated by artificial intelligence*, published February 28, 2025 and in force on the same day as the measures.[^4][^2] Labels were not new in 2025. The Provisions on the Administration of Deep Synthesis of Internet Information Services (CAC Order No. 12), in force since January 10, 2023, already required deep synthesis services to label what they generate or edit, and to label prominently anything that may confuse or mislead the public.[^5] The 2025 measures spell out who does what.

| Who | Duty | Article |
|---|---|---|
| AI service providers (the tools) | Add explicit labels where the service falls under Article 17(1) of the deep synthesis provisions; add implicit labels to file metadata; keep the explicit label in downloads and exports; explain the label format in the user agreement | Art. 4, 5, 8 |
| Content platforms | Check metadata and show one of three notices; provide a labeling tool and remind users to declare AI content; write the platform's own name or code and a content ID into the metadata | Art. 6 |
| App stores | Ask apps at listing whether they offer AI generation, and check their labeling materials if they do | Art. 7 |
| Users who post content | Declare AI content and label it with the platform's tool | Art. 10(1) |
| Anyone | Never maliciously remove, alter, forge or hide labels, supply tools for doing so, or use improper labeling to harm others | Art. 10(2) |

Source: full text of the measures.[^1] Breaches are handled by the cyberspace, telecom, public security and broadcasting authorities under existing laws and regulations (Art. 13).[^1]

The rules arrived in stages, shown in the timeline and table below.

![Timeline of China's AI content labeling rules: deep synthesis provisions in force January 2023, labeling measures and GB 45438-2025 in force September 1, 2025, livestream e-commerce measures in force February 2026](https://www.mvpdata.net/figures/ai-generated-content-labels/ai-labeling-rules-timeline-en.webp)

*Chinese rules and platform terms on AI content labels from 2023 to 2026, by effective or update date. Sources: original texts on cac.gov.cn, gov.cn and the national standards platform, and the Douyin and Xiaohongshu user agreements, read October 17, 2026.*

| Date | Event | Source |
|---|---|---|
| Jan 10, 2023 | Deep synthesis provisions (CAC Order No. 12) take effect | Art. 25[^5] |
| Feb 28, 2025 | Mandatory national standard GB 45438-2025 published | Standard record[^4] |
| Mar 14, 2025 | Four departments release the labeling measures (signed March 7, 2025) | CAC[^2], gov.cn[^1] |
| Sep 1, 2025 | Measures and GB 45438-2025 take effect together | Art. 14,[^3] standard record[^4] |
| Feb 1, 2026 | Livestream e-commerce measures (Order No. 117) take effect; Article 37 covers AI-generated people in live selling | Art. 69[^6] |
| Feb 20, 2026 | Current Douyin user agreement in force; clause 4.5 requires labels on non-real AI audio and video | Agreement page[^7] |
| Mar 23, 2026 | Xiaohongshu user agreement updated; clause 3.2 sets out "AI labels" | Agreement page[^8] |

## What is the difference between explicit and implicit AI labels?

An explicit label is a notice users can see or hear, in text, sound or graphics; an implicit label is data written into the file by technical means that users will not easily notice.[^1] When the four departments released the measures, they described the purpose of labels as alerting the public and supporting supervision and tracing.[^2]

![Explicit vs implicit AI content labels under China's 2025 measures: visible text, sound or graphic notices versus file metadata, and the three platform notices: is, may be, or is suspected to be AI-generated](https://www.mvpdata.net/figures/ai-generated-content-labels/explicit-vs-implicit-ai-labels-en.webp)

*Where explicit labels go, what implicit labels record and how platforms flag AI content once it is posted, from Articles 3 to 6 and 10 of China's labeling measures. Source: full text of the measures on gov.cn, read October 17, 2026.*

Article 4 sets where the explicit label goes for each type of content:[^1]

| Content | Where the explicit label goes |
|---|---|
| Text | A text notice or common symbol at the start, end or a suitable point in the middle, or a prominent notice in the interface or around the text |
| Audio | A spoken or rhythmic cue at the start, end or middle, or a prominent notice in the interface |
| Images | A prominent notice in a suitable position |
| Video | On the opening frame and in a suitable spot around the player; it may also appear at the end and in the middle |
| Virtual scenes | On the opening screen; it may also appear during the experience |

The implicit label goes into the file metadata, the descriptive information embedded in the file header. It records that the content is AI-generated, the provider's name or code, and a content ID. The measures also encourage digital watermarks as an additional implicit label (Art. 5).[^1]

Once content is posted, the platform shows one of three notices next to it (Art. 6):[^1]

| Situation | Notice the platform shows |
|---|---|
| The metadata states the content is AI-generated | The content is AI-generated |
| No implicit label, but the user declares it | The content may be AI-generated |
| No implicit label and no declaration, but the platform detects an explicit label or other traces | The content is suspected to be AI-generated |

In all three cases the platform also writes its own name or code and a content ID into the metadata.[^1] An AI video therefore carries a record from the tool and from the platform by the time it reaches viewers.

## What must creators and brands do when they post AI content?

Posters have two duties in China: declare AI content when publishing it and label it with the platform's tool, and leave intact any label the AI tool has already added.[^1] Creators, company accounts and brands are all "users" under Article 10, with the same obligations.

Which AI content must carry a visible label follows from Article 17(1) of the deep synthesis provisions. Services that may confuse or mislead the public must label prominently when they offer:[^5]

1. intelligent dialogue, intelligent writing and other text generation that imitates a person;
2. synthetic voices, voice imitation, or edits that significantly change a person's identifying features;
3. face generation, face swapping, face or pose manipulation, and other generation of images or video of people;
4. immersive, realistic virtual scenes;
5. any other service that generates content or significantly changes it.

Deep synthesis services outside these five must still offer a prominent labeling function and tell users they can use it.[^5]

Production teams often export finished files from professional tools. Under Article 9, a provider may supply content without the explicit label only after the user agreement sets out the user's labeling duty and responsibility, and it keeps a log of who received the file for at least six months.[^1] Once you hold a file without a visible label, labeling it is your job when you publish.

A pre-publication check for AI content:

- [ ] List every AI step in the piece: copy, voice, faces, footage, background music.
- [ ] Turn on the platform's AI declaration when posting (rules in the next two sections).
- [ ] Keep the original export from the AI tool and leave its watermark and metadata in place.[^3]
- [ ] If a real person's face or voice appears, confirm you have their consent; see [AI likeness rules since 2024](https://www.mvpdata.net/en/insights/ai-likeness-rules-timeline) and [digital replica](https://www.mvpdata.net/en/glossary/digital-replica).
- [ ] If the piece is sponsored or an ad, add the ad disclosure as well; see the section on ads and livestream selling below.

## How do Douyin, Xiaohongshu and WeChat Channels label AI content?

All three write AI labeling into their user agreements or operating rules: non-real content made with AI must be labeled prominently, and the platform may add labels of its own.[^7][^8][^9] As of October 17, 2026:

| Platform | Rule | What the poster must do | What the platform does | If you don't label |
|---|---|---|---|---|
| Douyin | User Service Agreement, clause 4.5 (version in force February 20, 2026) | Label prominently any non-real audio or video made with deep learning, VR, generative AI or similar technology, and other content that may confuse or mislead | May add notices around the content as the law requires | Not stated in this clause; the agreement refers to Douyin's announcement on upgrading its AI label feature |
| Xiaohongshu (RedNote) | User Service Agreement, clause 3.2 (updated March 23, 2026) | Label non-real content made with these technologies prominently; never present synthetic content as real | May add explicit and implicit labels, together called "AI labels"; users may not remove, alter, forge or hide them | Labels added, restrictions or bans; the user bears legal liability for the consequences |
| WeChat Channels | Operating Rules, clause 6.4 | Label prominently AI-generated or synthesized non-real audio and video, and fiction presented through staged plots | May add explicit and implicit labels and prominent notices | The platform may penalize unlabeled content, or such content used to gain traffic for improper marketing (fake staged footage, for example) |

Sources: Douyin user agreement,[^7] Xiaohongshu user agreement,[^8] WeChat Channels operating rules.[^9]

WeChat Channels extends the duty to staged fiction, a wider net than AI alone.[^9] Xiaohongshu spells out the consequences: added labels, restrictions and bans.[^8] Douyin leaves the detail to an in-app announcement;[^7] we found no public official copy, so we do not repeat press accounts of it.

Foreign brands posting AI assets on these three platforms follow the same rules; account set-up for China is covered on [China market entry](https://www.mvpdata.net/en/brands/china-entry).

## What are the AI label rules on TikTok, YouTube and Meta?

All three require labels on realistic AI content and may label automatically when a file carries C2PA metadata or similar signals.[^10][^11][^12] They differ on what counts as realistic, what the label says and whether an automatic label can be undone. As of October 17, 2026:

| Platform | What must be labeled | How to label | Automatic labels | If you don't label |
|---|---|---|---|---|
| TikTok | AI-generated content with realistic images, audio or video; labels are also encouraged for content fully generated or significantly edited by AI, such as people shown doing or saying what they did not, or face swaps | Turn on the "AI-generated content" setting before posting; the post shows "creator labeled as AI-generated" | "AI-generated" may be added for TikTok AI effects or uploads with C2PA Content Credentials; the creator cannot remove an automatic label | Misleading or unlabeled AI content that breaks the Community Guidelines may be removed |
| YouTube | AI-made or AI-altered content that seems realistic: a real person saying or doing something they did not, altered footage of real events or places, realistic scenes that never happened, and AI music that is the focus of the video | In YouTube Studio, under Attributes, answer Yes to "AI use" when uploading | Content made with YouTube's AI tools, content with C2PA metadata, and content its systems detect | Creators who consistently fail to disclose may get a manual label, removal, or suspension from the YouTube Partner Program |
| Meta (Facebook, Instagram, Threads) | Photorealistic video or realistic-sounding audio that was digitally created or altered may require an "AI content" label; the requirement does not apply to images | Add the "AI content" label when sharing; accounts that mostly post AI content can add an "AI creator" profile label | "AI content" when AI signals are detected; content made with Meta's AI tools may be labeled or watermarked | Penalties are possible when a required label is missing |

Sources: TikTok Help Center,[^10] YouTube Help,[^11] Meta Help Center.[^12]

The figure below compares all six platforms.

![AI-generated content label rules compared across Douyin, Xiaohongshu, WeChat Channels, TikTok, YouTube and Meta: what posters must do, when each platform labels automatically, and the consequences of not labeling](https://www.mvpdata.net/figures/ai-generated-content-labels/ai-label-rules-by-platform-en.webp)

*The AI label rules of six platforms, from their user agreements, operating rules and help centers. Sources: Douyin and Xiaohongshu user agreements, WeChat Channels operating rules, TikTok, YouTube and Meta help pages, read October 17, 2026.*

Three points for content teams:

- **The label itself does not cost reach.** TikTok says turning on the AI-generated content setting won't affect a video's distribution as long as it follows the Community Guidelines;[^10] YouTube says disclosing AI content alone won't limit a video's audience or its eligibility to earn money.[^11]
- **Light edits need no label on YouTube.** Beauty filters, color and lighting adjustments, captions, sharpening and upscaling, AI help with scripts and titles, and cloning your own voice for voice-overs are all on YouTube's no-disclosure list.[^11]
- **Some AI content is banned even with a label.** TikTok does not allow AI content that fakes authoritative sources or crisis events, falsely shows public figures making or receiving an endorsement, or uses the likeness of minors, or of adult private individuals without permission.[^10]

Chinese brands marketing overseas should build the declaration step into the posting schedule for these platforms; campaign planning is on [Social ads for Chinese brands going global](https://www.mvpdata.net/en/brands/go-global).

## What extra rules apply to AI in ads and livestream selling?

An AI label and an ad disclosure do different jobs, and a sponsored video made with AI needs both. How a sponsored post is disclosed in China, the US and Canada, including the hashtag Ad Standards lists for AI-made images in Canada, is set out in the [disclosure table on our operations page](https://www.mvpdata.net/en/services/operations#disclosure).

Livestream selling has its own rules. China's Measures for the Supervision and Administration of Livestream E-commerce (Order No. 117), in force since February 1, 2026, address AI directly:[^6]

| Article | What it says |
|---|---|
| Art. 17 | Livestream platforms must prevent and deal with room operators and hosts who use AI to fabricate or spread false or misleading commercial information or to impersonate others |
| Art. 34(2) | Room operators and hosts may not use AI to fabricate or spread false or misleading commercial information, or to impersonate others in promotion |
| Art. 37 | A room operator that uses AI-generated images or video of people in live selling must meet the laws, regulations and mandatory national standards, label them, and keep telling consumers throughout that they are AI-generated; the operator that manages or uses them is liable for any breach |
| Art. 59(5) | Failing the Article 37(1) labeling duty, where laws and administrative regulations are silent: an order to correct, then a fine of 10,000 to 100,000 yuan if the operator refuses or the case is serious |
| Art. 65(2) | Where the cyberspace authority is responsible, breaches of Article 37 and others are penalized first under the Cybersecurity Law and other laws and administrative regulations; where they are silent, an order to correct, then a fine of 10,000 to 100,000 yuan and possible suspension of services if the operator refuses or the case is serious |

The key word in Article 37 is "continuously": for as long as an AI-generated person appears in the stream, shoppers should be able to see that it is AI-generated.

Overseas, ad rules also keep the AI label separate from the ad disclosure. Meta adds an "AI info" label to ad images created or significantly edited with its generative AI ad features or with third-party AI tools; it appears on the "About this ad" screen in the ad's three-dot menu and sometimes next to "Sponsored", and it is not added for minor changes such as resizing or color correction.[^13]

## Which common claims about AI labels are out of date?

Several claims that appear in search results and AI answers do not match the current texts. Each line below was checked against the original on October 17, 2026:

| Claim | What the source says | Source |
|---|---|---|
| "China's labeling rules only apply to AI companies, not creators" | Users who post AI content must declare it and label it with the platform's tool (Art. 10(1)) | Labeling measures[^1] |
| "A watermark is enough" | There are two kinds of label; a watermark is one form of implicit label the measures encourage, while the implicit label in the metadata is required | Labeling measures[^1] |
| "The CAC issued the measures on its own" | Four departments issued them jointly, and a mandatory national standard, GB 45438-2025, applies alongside | CAC release,[^2] standard record[^4] |
| "On YouTube you disclose under an Altered content setting" | The help page now points to "AI use" under Attributes at upload | YouTube Help[^11] |
| "Meta's AI label is called AI info" | Organic posts are labeled "AI content"; "AI info" is the label on ad images | Meta Help Center[^12][^13] |
| "AI images must be labeled on Meta" | The requirement covers photorealistic video and realistic audio, not images, though images can still be labeled when AI signals are detected | Meta Help Center[^12] |
| "Labeling AI content kills your reach" | TikTok: the setting does not affect distribution if the post follows the guidelines; YouTube: it does not limit audience or monetization | TikTok,[^10] YouTube[^11] |

## How does MVP check AI content before it goes out?

MVP reviews every post before it goes live as part of its channel operations: the team checks facts, copyright and sensitive wording, and marks sponsored content as advertising, as described on [channel operations](https://www.mvpdata.net/en/services/operations). For AI-assisted content, the labeling rules apply on top of the ad disclosure, and the two platform tables above work as a checklist.

MVP (Mega Volume Production, 十方众声) is headquartered in Toronto and contracts through two companies, 北京十方众声传媒有限公司 and Mega Volume Production Inc., working in talent management and cross-border brand marketing. On the brand side that covers [social ads for Chinese brands going global](https://www.mvpdata.net/en/brands/go-global), [China market entry](https://www.mvpdata.net/en/brands/china-entry) and [cross-border influencer marketing](https://www.mvpdata.net/en/brands/influencer-marketing); when a client's face or voice is copied with AI, see [IP protection](https://www.mvpdata.net/en/services/protection). To discuss a campaign or a content plan, [contact MVP](https://www.mvpdata.net/en/contact).

This article summarizes public rules and is not legal advice. Platform rules change; the current version on each platform's own page prevails.

## Questions

### Do I need to label AI-written captions and scripts?

It depends on where the text goes. China's measures treat AI-generated text as synthetic content that posters must declare,[^1] and AI chat and writing services that imitate a person must add a visible label.[^5] YouTube, by contrast, says using AI for an outline, script or title needs no disclosure.[^11]

### Can I remove an AI tool's watermark before posting in China?

Not as a way around the rules. Article 10 of the measures bars anyone from maliciously removing, altering, forging or hiding labels, or supplying tools to do so.[^3] A file without a visible label can be requested from the tool provider under Article 9, but only after the user agreement assigns you the labeling duty; the provider keeps a log of who received it for at least six months.[^1]

### Do beauty filters and color correction count as AI content?

On the overseas platforms, no: YouTube lists beauty filters, color adjustment and lighting filters among edits that need no disclosure,[^11] and Meta does not label ad images that were only resized or color-corrected.[^13] China's measures give no list of exempt edits; the test is whether the content was generated or synthesized with AI.

### What if a platform labels my real footage as AI-generated?

On YouTube you can usually switch the AI disclosure to No under Attributes in YouTube Studio, except for content made with YouTube's own AI tools, content with C2PA metadata, or labels applied after manual review.[^11] On TikTok, an automatic AI-generated label cannot be removed by the creator.[^10]

## Related reading

- [China social media management for public figures and brands](https://www.mvpdata.net/en/services/operations): MVP (Mega Volume Production, 十方众声) provides China social media management for established creators, celebrities, founders and companies. Its in-house team plans content, schedules posts, looks after fan communities, reviews the numbers and checks every post before it goes live, on Douyin, Xiaohongshu, Weibo, Bilibili and WeChat as well as YouTube, Instagram and TikTok. The accounts stay in your name.
- [China entry: PR and social media ads for foreign brands](https://www.mvpdata.net/en/brands/china-entry): MVP (Mega Volume Production, 十方众声) helps foreign brands enter China. It distributes Chinese-language press releases and works with the media, runs ads on Douyin, Xiaohongshu (RedNote), WeChat and Channels, Weibo, Bilibili, Kuaishou and Baidu, sets up and runs official Chinese accounts, and connects brands with Chinese creators. One team runs all four, in ad accounts opened in MVP's name, with reach and ROI targets written into the contract.
- [Social media advertising for Chinese brands going global](https://www.mvpdata.net/en/brands/go-global): MVP (Mega Volume Production, 十方众声) helps Chinese companies and brands grow abroad, mainly in North America: it runs ads on TikTok, Facebook and Instagram, Google and YouTube, X and LinkedIn, manages official accounts, brings in overseas creators and distributes English press releases. Ad accounts are opened in MVP’s name and MVP settles the spend, and reach and ROI targets can be written into the contract.
- [Celebrity reputation management and risk protection](https://www.mvpdata.net/en/services/protection): MVP (Mega Volume Production, 十方众声) provides celebrity reputation management and crisis response for celebrities, creators, founders and companies. It watches for risks, and when something goes wrong it checks the facts, prepares the response, reports false or infringing content and helps repair the damage. Public statements go out in your name, never MVP's. Long-term work includes monitoring and crisis response; a single incident can be a one-off project.
- [AI likeness and voice rules in China and the US since 2024](https://www.mvpdata.net/en/insights/ai-likeness-rules-timeline): As of October 2026, AI likeness and voice rules in China and the US have moved three ways since 2024: the law now expressly protects the voice, including AI-processed voices; regulators require AI-generated content to carry labels; and platforms have opened likeness detection to talent and their agencies. This article lists the six changes in date order, each with its official source.

[^1]: 《关于印发《人工智能生成合成内容标识办法》的通知》 (Notice on Issuing the Measures for Labeling AI-Generated and Synthetic Content), Cyberspace Administration of China, Ministry of Industry and Information Technology, Ministry of Public Security, National Radio and Television Administration (gov.cn), 国信办通字〔2025〕2 号. Effective 2025-09-01. https://www.gov.cn/zhengce/zhengceku/202503/content_7014286.htm (Checked 2026-10-17, Unofficial translation)
[^2]: 《四部门联合发布《人工智能生成合成内容标识办法》》 (Four departments jointly issue the Measures for Labeling AI-Generated and Synthetic Content), Cyberspace Administration of China (cac.gov.cn). Published 2025-03-14. https://www.cac.gov.cn/2025-03/14/c_1743654685899683.htm (Checked 2026-10-17, Unofficial translation)
[^3]: 《人工智能生成合成内容标识办法》 (Measures for Labeling AI-Generated and Synthetic Content), Cyberspace Administration of China and three other departments, 国信办通字〔2025〕2 号. Effective 2025-09-01. https://www.cac.gov.cn/2025-03/14/c_1743654684782215.htm (Checked 2026-10-17, Unofficial translation)
[^4]: 《网络安全技术 人工智能生成合成内容标识方法》 (Cybersecurity technology: Labeling method for content generated by artificial intelligence), SAMR and Standardization Administration of China (national standards information platform), GB 45438-2025. Effective 2025-09-01. https://std.samr.gov.cn/gb/search/gbDetailed?id=301E0388CB75788DE06397BE0A0AE1B4 (Checked 2026-10-17)
[^5]: 《互联网信息服务深度合成管理规定》 (Provisions on the Administration of Deep Synthesis of Internet Information Services), Cyberspace Administration of China, Ministry of Industry and Information Technology, Ministry of Public Security, 令第 12 号. Effective 2023-01-10. https://www.cac.gov.cn/2022-12/11/c_1672221949354811.htm (Checked 2026-10-17, Unofficial translation)
[^6]: 《直播电商监督管理办法》 (Measures for the Supervision and Administration of Livestream E-commerce), Cyberspace Administration of China (cac.gov.cn), 令第 117 号. Effective 2026-02-01. https://www.cac.gov.cn/2026-01/07/c_1769516655383334.htm (Checked 2026-10-17, Unofficial translation)
[^7]: 《“抖音”用户服务协议》 (Douyin User Service Agreement), Douyin. Effective 2026-02-20. https://www.douyin.com/agreements/?id=6773906068725565448 (Checked 2026-10-17, Unofficial translation)
[^8]: 《小红书用户服务协议》 (Xiaohongshu User Service Agreement), Xiaohongshu (RedNote). Page updated 2026-03-23. https://agree.xiaohongshu.com/h5/terms/ZXXY20220331001/-1 (Checked 2026-10-17, Unofficial translation)
[^9]: 《微信视频号运营规范》 (WeChat Channels Operating Rules), Tencent WeChat. https://weixin.qq.com/cgi-bin/readtemplate?lang=zh_CN&t=weixin_agreement&s=video (Checked 2026-10-17, Unofficial translation)
[^10]: About AI-generated content, TikTok Help Center. https://support.tiktok.com/en/using-tiktok/creating-videos/ai-generated-content (Checked 2026-10-17, May be unavailable in mainland China)
[^11]: Disclosing use of GenAI content, YouTube Help. https://support.google.com/youtube/answer/14328491 (Checked 2026-10-17, May be unavailable in mainland China)
[^12]: How to identify AI content on Meta products, Meta Help Center. https://www.meta.com/help/artificial-intelligence/1783222608822690/ (Checked 2026-10-17, May be unavailable in mainland China)
[^13]: How AI-generated images in ads are identified and labeled on Meta, Meta Help Center. https://www.meta.com/help/artificial-intelligence/355108217670024/ (Checked 2026-10-17, May be unavailable in mainland China)