---
title: "YouTube 將 AI 深度偽造檢測工具擴展至更多創作者：這是什麼"
type: "News"
locale: "zh-HK"
url: "https://longbridge.com/zh-HK/news/286734835.md"
description: "YouTube 正在將其基於人工智能的肖像識別工具擴展到更多創作者，以打擊深度偽造和人工智能生成的冒充行為。該功能最初僅對部分創作者開放，現在將向所有符合條件的 18 歲以上創作者提供。它允許創作者識別並請求刪除未經授權的 AI 修改視頻。這一舉措是在對深度偽造內容日益擔憂的背景下推出的，近年來此類內容激增了 900%，尤其對女性影響顯著。該工具旨在增強創作者的控制權，並減少平台上的錯誤信息"
datetime: "2026-05-17T23:55:04.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/286734835.md)
  - [en](https://longbridge.com/en/news/286734835.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/286734835.md)
---

# YouTube 將 AI 深度偽造檢測工具擴展至更多創作者：這是什麼

YouTube is expanding its AI-powered likeness detection feature to more creators as concerns around deepfakes and AI-generated impersonation continue to grow. According to the company’s support page, the feature is designed to help creators detect videos where their face may have been altered or generated using AI and request removal if the content violates YouTube’s privacy rules.

The tool, previously limited to select creators and public figures, will now gradually roll out to all eligible creators above the age of 18 in the coming weeks.

The announcement comes as AI-generated videos become easier to create and more widely shared online. Platforms such as YouTube are increasingly facing pressure to provide tools that help users protect their identity and prevent misleading content.

**What is YouTube’s likeness detection tool?**

Likeness detection is a feature inside YouTube Studio that scans videos on the platform to identify possible AI-generated or altered versions of a creator’s face. According to YouTube, the tool is intended to give creators more control over how their likeness is used online.

The feature first launched in October 2025 for a limited group of creators in the YouTube Partner Program. Over time, the company expanded access to government officials, politicians, journalists, and entertainment professionals. Now, YouTube is making it available to a wider creator base.

YouTube’s “likeness” feature refers to AI tools that let creators use a digital AI version of themselves for making Shorts and other content. The feature can generate videos using a creator’s face, voice, and appearance without them needing to film every clip manually.

The detection tool available in YouTube Studio uses a one-time facial verification process to identify “altered or synthetic” uses of a person’s likeness across YouTube. Creators can then review flagged videos and request removal directly through YouTube Studio if they believe the content is unauthorised.

The feature works inside YouTube Studio and scans uploaded videos for possible AI-generated or altered versions of a creator’s face. If a match is detected, creators can review the flagged content and request removal directly through the platform.

As of March 2026, studies have pointed to a sharp rise in AI-generated deepfake content online, according to UK-based cybersecurity intelligence firm Pi-Labs, reported Business Standard. The report noted a 900 per cent increase in deepfake material in recent years and found that more than 90 per cent of explicit deepfakes target women.

The misuse of generative AI is increasingly linked to non-consensual sexual imagery, identity manipulation, and image morphing. In India, the issue is also reflected in cybercrime data, with complaints involving women reportedly rising from around 50,000 in 2024 to nearly 80,000 by 2026, marking a 60 per cent increase in two years.

YouTube explained that the system is designed to help creators track where their faces may be appearing in AI-generated videos uploaded by others.

The feature also allows creators to:

-   Identify videos using AI-generated versions of their face
-   Request removal of fake or manipulated content
-   Help prevent viewers from being misled by impersonation videos

**What creators can do with it**

According to the company, the tool allows creators to check whether their likeness is being used in AI-generated videos uploaded by others. It can also help reduce misinformation by making it easier to remove misleading content that falsely appears to feature them.

YouTube said the feature is aimed at protecting both creators and audiences, especially as AI-generated impersonation becomes more convincing.

The company added that creators may not immediately see detected matches after enabling likeness detection. According to YouTube, this could simply mean that AI-generated or altered videos featuring their face are not being uploaded frequently. The platform said the tool will continue running in the background to monitor and help detect unauthorised use of a creator’s likeness.

**ALSO READ: Instagram 'Instants': Step-by-step guide to hide the feature, unsend photos**

**How to enable the feature**

Open YouTube Studio on your device

From the left menu, select Content Detection, then Likeness, and tap Start now

Permit YouTube to use likeness detection technology

Complete the one-time verification process to enable the feature

**ALSO READ: OnePlus halts OxygenOS 16 update rollout over boot-related problems**

**Other platforms**

As generative AI tools become capable of producing realistic text, images, audio, and videos, tech companies are increasingly focusing on ways to identify whether online content was created using AI.

Google introduced SynthID Detector, a verification platform designed to detect AI-generated content created through its AI models. The system uses invisible watermarks embedded into AI-generated media without affecting quality and can identify AI-created text, images, audio, and videos generated using models such as Gemini, Imagen, Lyria, and Veo.

Adobe, meanwhile, uses Content Credentials, an industry-standard metadata label that shows how digital content was created or edited. It can include details such as the creator’s identity, whether the content was captured using a camera, generated through AI, or edited with tools such as Photoshop.

LinkedIn also uses Content Credentials, a system built on the C2PA standard, to show the origin and editing history of digital content. The metadata can indicate whether content was edited, AI-generated, or captured directly from a camera.

### 相關股票

- [XSW.US](https://longbridge.com/zh-HK/quote/XSW.US.md)
- [GGLL.US](https://longbridge.com/zh-HK/quote/GGLL.US.md)
- [XDAT.US](https://longbridge.com/zh-HK/quote/XDAT.US.md)
- [IGV.US](https://longbridge.com/zh-HK/quote/IGV.US.md)
- [GOOGL.US](https://longbridge.com/zh-HK/quote/GOOGL.US.md)
- [GOOG.US](https://longbridge.com/zh-HK/quote/GOOG.US.md)

## 相關資訊與研究

- [第三方帳號成資安破口，SailPoint：很多企業連「誰能登入」都不知道](https://longbridge.com/zh-HK/news/286868553.md)
- [Siri 加入「聊天自動刪除」功能？蘋果再靠「隱私」魔法打 AI 戰](https://longbridge.com/zh-HK/news/286703500.md)
- [【行業透視】善用 AI 重在定義問題與專業把關](https://longbridge.com/zh-HK/news/286681677.md)
- [數發部發布 AI 產業人才認定指引 3.0 擴大生態系](https://longbridge.com/zh-HK/news/286858820.md)
- [圓剛法説會／深化邊緣 AI 與創作者生態佈局 雙引擎推下階段成長動能](https://longbridge.com/zh-HK/news/286538207.md)