What is a deepfake?
Images and audio can be edited without consent and used for impersonation, harassment, fraud, sexual image abuse, and credibility attacks.
A deepfake is a technology, or a generated result, that uses AI or similar methods to synthesize or edit a person's face, voice, or movements so they appear real.
This technology is also used in video production and creative work.
On the other hand, it is also used for impersonation, harassment, fraud, sexualized image abuse, and reputational harm using someone's face or voice without consent.
When thinking about anonymity, the important point is that images and audio you publish can later be used in another context.
This article explains the basics of deepfakes and the points ordinary individuals should watch for.
What can be made with deepfakes?
Deepfakes can be used for face replacement, voice synthesis, generating mouth movements, creating fictional people, and similar actions.
| Type | Description |
|---|---|
| Face synthesis | Overlaying another person's face on video or an image |
| Voice synthesis | Creating speech in a voice similar to the person's |
| Lip-sync generation | Creating mouth movements matched to audio |
| Fictional person generation | Creating an image of a person who does not exist |
| Image alteration | Making it look as if the person did something they did not do |
The technology itself is not all bad.
The problem is when it is used without the person's consent or to deceive viewers.
Published material can be used
Deepfakes need source material.
Face photos, videos, voices, social media posts, streams, profile images, family photos. The more public material exists, the more material there is for misuse.
Voice also becomes personal information
In deepfakes, voice is important source material as well as the face.
If audio streams, videos, call recordings, voice posts on social media, or event remarks remain, they may be used for voice synthesis or impersonation.
| Audio material | Risk |
|---|---|
| Stream archives | Long audio remains |
| Voice in video | Face and voice are paired |
| Voice posts | Voice alone can sound recognizable |
| Event remarks | Links with activity or affiliation |
| Call recordings | Can be used to impersonate identity verification |
Even without showing a face, voice becomes a clue that can make someone recognizable.
| Material | Risk |
|---|---|
| Face photo | Used for face synthesis or impersonation |
| Video | Becomes material for facial expressions and angles |
| Audio | Used for voice synthesis |
| Family photo | People around the person may be pulled in too |
| Past posts | Become material for context and realistic impersonation |
This is not as simple as saying you are completely safe if you never show your face or voice.
However, the less public material there is, the less material there is for misuse.
Relationship to anonymity
Deepfakes also relate to anonymity.
If someone's real face or voice is misused, anonymous activity may be linked with real-name information, or fake information may be used to damage credibility.
| Risk | Impact on anonymity |
|---|---|
| Impersonation | Makes it look as if the person made a statement |
| Fake image | Links anonymous activity with the person's real-name identity or face |
| Voice misuse | Makes phone calls or voice messages seem like the person |
| Harassment | Threatens social credibility and safety |
| Pulling in people involved | Family or allies' images are also used |
In anonymous activity, decide carefully whether to publish face or voice.
Once material spreads, it cannot be fully recovered later.
Public materials to watch
As a deepfake countermeasure, deleting every photo and audio file is not realistic.
However, you can be careful before publishing.
| Published item | What to check |
|---|---|
| Face photo | Whether you are publishing many high-resolution images |
| Video | Whether it contains enough face angles and voice |
| Audio stream | Whether voice alone makes the person recognizable |
| Family photo | Whether you considered consent and impact on people around you |
| Profile image | Whether it is reused between real-name and anonymous activity |
Videos that leave face and voice together become especially strong material.
Think about whether publication is necessary and what the risks are.
If you upload face photos, voices, victim images, or suspicious videos to external generative AI, voice-cloning, image-editing, or deepfake detection services, the material itself, source IP address, usage time, browser information, and search or detection history may be handed to that service. For high-risk material or victim images, it is important not to casually hand them to external services even for checking.
Judging real from fake becomes harder
The problem with deepfakes is not only that fakes can be created.
Even real images and audio may be called "possibly fake." This can force victims to explain themselves and damage their credibility.
| What happens | Impact |
|---|---|
| Fakes are treated as real | Leads to impersonation or harassment |
| Real material is treated as fake | Credibility of testimony and records is shaken |
| People are forced to explain | The burden shifts to the victim |
| Spread moves ahead first | Corrections are less likely to reach people |
| People involved are pulled in | Family and workplaces may be affected |
That is why you need to manage published material and think about evidence preservation if a problem occurs.
If you notice misuse
If you notice a deepfake or impersonation, record it before reacting in a panic.
Save the URL, screenshots, posting time, account, and spread status. If sexualized images, threats, minors, fraud, or defamation are involved, consider platform reporting, specialists, lawyers, police consultation, and support organizations.
| Situation | Response |
|---|---|
| Impersonation post | Save the URL, screenshot, and account |
| Misuse of sexualized images | Quickly consider reporting and specialized support channels |
| Use in fraud | Warn people involved and keep evidence |
| Threat | Prioritize safety and consultation over replying |
| Spread | Request removal and check the scope of impact |
It is important not to handle it alone.
Everyday prevention
It is difficult to completely prevent deepfakes.
Even so, you can reduce the material available for misuse by limiting public material, not reusing images or voices between real-name and anonymous activity, and not publishing photos of family or allies without permission.
| Prevention | Effect |
|---|---|
| Do not publish too many face photos | Reduces the amount of source material |
| Be careful with video publication | Reduces face-and-voice pairs |
| Separate images between real-name and anonymous use | Avoids correlation |
| Do not overpublish family photos | Avoids pulling in people around you |
| Decide how to record problems if they occur | Makes consultation and reporting easier |
What matters is not becoming so afraid that you can do nothing.
It is choosing for yourself which material to publish.
Watch material involving family and allies
Deepfake harm does not necessarily happen only to you.
Family photos, group photos with allies, videos during activity, children's photos, and workplace or school photos can also become material for misuse. If you publish material without the person's consent, you expose that person to risk.
| Material | Caution |
|---|---|
| Family photo | Faces and regular locations appear |
| Child photo | The risk of remaining into the future is large |
| Group photo with allies | Easily becomes a list of people involved |
| Workplace or school photo | Affiliation and position become visible |
| Activity video | Face, voice, and claims remain together |
Even for material you want to publish, handle it carefully if it includes another person's face or voice.
Do not overexplain after harm occurs
When you notice misuse, you may want to immediately write a long explanation.
However, explanations may reveal new personal information. First preserve evidence and consider reporting or consultation options.
| Common rushed action | Caution |
|---|---|
| Write a long rebuttal | Easy to reveal additional personal information |
| Chase the other party | May lead to further harassment |
| Delete evidence | Makes consultation and reporting harder |
| Spread it to people involved | May further spread victim images |
| Carry it alone | Delays judgment |
After harm occurs, prioritize preservation, reporting, and consultation before public rebuttal.
Summary
A deepfake is a technology or generated result that uses AI or similar methods to synthesize or edit a face or voice so it appears real.
While it can be used for creative work, it is also used for impersonation, harassment, fraud, and misuse of sexualized images.
Published face photos, videos, audio, and profile images may later be used in another context.
When thinking about anonymity, check how much of your face and voice to publish and whether images are being reused between real-name activity and anonymous activity.
If you notice misuse, preserve evidence and consider reporting or consulting specialists as needed.
Related tools
Google Lens
An external resource related to this article. Open it only when it fits your situation and threat model.
Why it is listed: It can help with the article topic, but it is outside Anonymity Sense and should be checked before use.
URL : https://lens.google/
PimEyes
An external resource related to this article. Open it only when it fits your situation and threat model.
Why it is listed: It can help with the article topic, but it is outside Anonymity Sense and should be checked before use.
URL : https://pimeyes.com/
ExifTool
An external resource related to this article. Open it only when it fits your situation and threat model.
Why it is listed: It can help with the article topic, but it is outside Anonymity Sense and should be checked before use.
URL : https://exiftool.org/
MAT2
An external resource related to this article. Open it only when it fits your situation and threat model.
Why it is listed: It can help with the article topic, but it is outside Anonymity Sense and should be checked before use.
FFmpeg
An external resource related to this article. Open it only when it fits your situation and threat model.
Why it is listed: It can help with the article topic, but it is outside Anonymity Sense and should be checked before use.
URL : https://ffmpeg.org/