- Tool to detect and manage deepfakes that use your face, with privacy or copyright actions from YouTube Studio.
- Available to eligible YPP creators; requires verification with official document and selfie video.
- Biometric data used only for detection; stored for up to 3 years and deletable upon withdrawal of consent.
- The review considers parody, satire, and AI disclosure; you can choose to archive, withdraw, or claim rights.
Finally, YouTube has a tool designed to protect your identity from deepfakes. Its name: YouTube Likeness DetectionIt's a solution similar to other platforms that apply Measures to reduce AI-generated contentWith it, creators can locate videos where your face has been altered or generated by AI and decide if they want to ask them to withdraw.
This technology works similarly to Content ID, but instead of searching for copyrighted audio or video matches, Track your facial resemblanceAfter you provide a reference image of your face during setup, the system analyzes new uploads to identify potential matches. It's in its early stages and is still improving, so you'll see both accurate matches and, occasionally, false positives; even so, It makes it easy to request withdrawals under the privacy policy. and provides a clear panel for reviewing cases.
What is Likeness Detection and what is it used for?
The tool detects videos in which Your face may have been manipulated or created with AIIf it finds results, it allows you to review them in YouTube Studio and choose what to do in each case. YouTube uses automated systems for many tasks (ad suitability, copyright, or abuse prevention) always in accordance with the Community Guidelines; in this context, Likeness Detection adds a layer to manage the use of your image to scale.
Important: You can only identify the resemblance of eligible creators who have given their consentIt is not designed to recognize other people who appear in videos uploaded to the platform, nor to monitor third parties outside the scope of those who activate the function.
Availability, eligibility and access
The deployment has begun with creators of the YouTube Partner Program (YPP) and will be expanded in the coming months. The first wave received an invitation email, and gradually more channels will see the tab enabled. During a pilot phase, YouTube collaborated with CAA (Creative Artists Agency) to validate the approach with artists, celebrities and creators exposed to deepfakes, and has showcased the feature on its Creator Insider channel.
To configure it you must have more than 18 years and be the Channel Owner or listed as a Manager; Editors can view and act on detected videos, but cannot create the initial video. Any delegate with access to the tab of Content detection is considered an authorized representative to raise a privacy complaint without additional verification (roles: Manager, Editor and Limited Editor).
How to get started step by step
You can start the YouTube Likeness Detection process from YouTube Studio. In the side menu, go to Content detection > Likeness and click on «Start now» to start the setup. Here you will need accept the use of biometric technology in order to find your lookalike on YouTube, something key to preventing fraud and abuse.
Onboarding includes mobile identity verification: scan the QR code displayed on the screen and complete the flow by uploading a photo of your official document and a brief selfie videoThat short recording, along with images of your face from your own YouTube content, is used to generate facial (and, in certain cases, voice) templates that serve to detect AI-altered appearances. Practical advice: Use a clear and legible photo of your document to avoid rejection.
After submitting the documentation, you will receive a confirmation email When everything is ready. This process can take up to 5 days from when you submit your ID/passport and the selfie video; keep this in mind if you need to use the tool urgently.
Review of matches and available actions
Once you have access, return to YouTube Studio and enter Content detection > Likeness > For reviewThere you will see the matches that the system has detected, with the option to filter by playback volume (Total views) or by channels ordered according to their number of subscribers (Subscribers), which helps you prioritize.
By pressing «ReviewNext to a video, a detailed view opens to help you decide if you think your image or your voice... have been altered or generated with AIIf you select "Yes", the system offers you two options: do nothing (leave the video as is) or request withdrawal If you believe that YouTube's use of your image/voice, infringing its privacy policy, please complete the form with the required information.
If you answer "No" (it's not altered by AI), the flow will ask for more context: you can indicate that it's about real material of yours or that That's not your faceIn which case, the item will be moved to the "Archived" tab. This option is useful for clearing the panel of matches that don't require action, allowing you to focus on what's most important.
Privacy vs. copyright: two different paths
In Likeness Detection, two regulatory frameworks with different criteria coexist. On the one hand, the Privacy Policy It addresses cases where your image is used in an altered or synthetic way to suggest actions, support, or messages that are not yours (for example, videos in which it appears that you support a candidate or infomercials that They show their faces to you without permission). On the other hand, the Copyright They refer to the use of your original content (clips from your videos, audio, etc.), with considerations of legitimate/fair use.
The tool may uncover actual clips of you that don't fit the privacy guidelines; in those cases, proposes a copyright withdrawal if applicable. YouTube also notes that it values factors such as parody or satire and if the video includes a AI usage statement when considering whether or not to remove content following a privacy complaint.
Panel and user experience
The YouTube Likeness Detection dashboard shows titles, upload date, and the channel that published it. view counts and subscribers, and can mark certain matches as «high prioritySo you can attend to them first. If you prefer, you can file a case when you are not going to take action and leave a record for future reference.
For channels experiencing large-scale deepfakes, the manual review process can be demanding. YouTube has acknowledged the challenge, and while the initial approach is case-by-case—similar in spirit to Content ID— The company is gathering feedback to evolve the tool and respond to scenarios with hundreds or thousands of forgeries.
Key FAQs
- Why can't I see detected videos? It's normal for this to affect you at the beginning, or if only a few fake videos are uploaded. An empty list suggests that no unauthorized use has been detected so far. If you discover a video that isn't listed, please report it using the privacy form for review.
- Why didn't the tool detect one of my deepfakes? The technology is in the experimental phase and is still being refined. You can submit a privacy takedown request via the form if something is leaking from the dashboard. For voice imitations, please use the same reporting channel.
- Who can do the configuration? Channel Owner or Managers. Editors have permission to view and act, but not to initiate the registration process.
- What if it's my real face in the video? Likeness may display snippets of your original content. These are not removed for privacy reasons, although you can file a copyright complaint if applicable and fair use does not apply.
- Who is authorized to file a privacy complaint? Anyone with a role that allows access to Content detection (Manager, Editor, Limited Editor) is considered an authorized representative and does not need additional verification.
How YouTube uses and stores your data
If you sign up, YouTube creates templates of your face (and can generate your voice from your own content) using the verification selfie video and screenshots from your videos. They are used to detect coincidences in altered or synthetic content where your image appears. Your full legal name, collected during verification, is used to comply with legal requirements in removal requests.
When several people in a channel have configured YouTube Likeness Detection, the system displays the legal name alongside the videos where each one appears, so that any authorized user of the channel can filter and review cases per person easily. Additionally, when processing a withdrawal, the YouTube operations team may see a selfie video capture to validate faster that you are who you say you are.
In storage, your selfie video, legal name, and templates are assigned a unique identifier and are stored in YouTube's internal databases for up to 3 years from your last access to YouTube, unless you withdraw your consent or delete your account. You can unsubscribe at any time from «Manage likeness detection"If you do this, this data will be deleted and..." Scanning of new videos ceasesYour official document data is stored in your Google Payments Profile, where you can access and delete it whenever you want.
Signing up for this feature does not grant YouTube permission to train generative models with your content beyond the specific purpose of similarity detection. YouTube does not store data on those who may appear in the scanned videos; that is, it does not create biometric databases of non-participating third parties.
Privacy complaint management
When you submit a privacy takedown request and the content is removed, you will receive an email with the result. YouTube tries to process these requests as quickly as possible; if you're worried about the time, Contact your partner manager If you have one. If you change your mind and want to withdraw your complaint, reply to the acknowledgment email to request the retraction.
Not all content is removed: YouTube considers factors such as parody, satire, and whether the video contains AI use disclosure or other criteria. The review seeks to balance the protection of identity with freedom of creation, attempting to avoid undue removals while addressing the malicious use of deepfakes.
Context: AI policies and other initiatives at YouTube
The platform demands label the content that have been generated or altered with AI under certain circumstances, especially if they could be misleading. In the music sector, it has announced a strict policy against voice imitations of artists. In addition, YouTube is experimenting with creative tools such as Dream Screen for Shorts, incorporating safeguards that block prompts that violate policies or touch on sensitive topics.
The company argues that AI should to enhance human creativitynot replace it. That's why it collaborates with partners and creators to create safeguards and reduce harmful uses, while also fostering responsible innovation. On the regulatory front, YouTube has expressed its support for NO FAKES Act, a US proposal to tackle the illegitimate use of image or voice for deceptive purposes.
Current limitations and realistic expectations
It's reasonable to assume that detection isn't perfect: there will be omissions and dubious coincidencesespecially with subtle manipulations. There's also the operational challenge of reviewing large volumes of results if you're a high-profile creator. Even so, having a single control panel The ability to escalate withdrawals, archive, or copyright claims represents a significant practical advance.
If you don't see any matches, don't worry; there may not be any. unauthorized uses detected or that you are in the early stages of deployment. On the other hand, if you detect a problematic video that is not appearing, the privacy form It remains the valid way for YouTube to evaluate it according to its rules.
What to expect from the ecosystem in the medium term
As AI creation becomes more widespread, we will see more sophistication in the impersonation techniques And, in parallel, improvements to defensive systems like YouTube Likeness Detection. The platform's objective is to give creators the tools to... maintain control about their digital identity, while also protecting legitimate expressions such as satire. The lessons learned in this initial phase—plus collaboration with labels, agencies, and artists—will shape the evolution of the feature and potential for further automation.
With YouTube Likeness Detection, YouTube puts a clear mechanism in the hands of creators to detect and manage deepfakes that use their image, with a robust verification process, an orderly review panel, and differentiated courses of action between privacy and copyright. Although there is still room for improvement—especially in scale and voice coverage—its progressive rollout, support for initiatives like NO FAKES, and AI policy safeguards paint a picture in which defend your identity It's simpler and, above all, faster.
Editor specialized in technology and internet issues with more than ten years of experience in different digital media. I have worked as an editor and content creator for e-commerce, communication, online marketing and advertising companies. I have also written on economics, finance and other sectors websites. My work is also my passion. Now, through my articles in Tecnobits, I try to explore all the news and new opportunities that the world of technology offers us every day to improve our lives.