In short: a blackmailer no longer needs you to have done anything at all. A few seconds of your face captured during a random video chat are now enough to build a faked image or short video — a deepfake, or hypertrucage (deepfake) — convincing enough to alarm a parent, an employer or a community. For an adult gay, bisexual or questioning man who is not out, the threat is twofold: the content is fake, but the outing it causes is real. This guide explains how to recognise a doctored image, how to prove it is one, how to get it taken down quickly, and exactly what French and Belgian law says — since 2024, France specifically punishes the distribution of sexual deepfakes without consent.
This article describes personal safety practices, public services and the general legal framework. It does not constitute personalised legal advice. In France: the plateforme PHAROS du ministère de l'Intérieur (PHAROS platform of the Ministry of the Interior) at internet-signalement.gouv.fr, 17 or 112 in an emergency, 114 by text message, the SOS homophobie (SOS Homophobia) helpline on 01 48 06 42 41 and its website chat, Cybermalveillance.gouv.fr, 3018 (the national platform against digital violence, free and anonymous), the CNIL (National Commission on Information Technology and Civil Liberties) for personal data, the Défenseur des droits (Defender of Rights) on 09 69 39 00 00, 116 006 (France Victimes / France Victims), and Arcom (French audiovisual and digital communication regulator) for reporting platforms. In Belgium: Unia on 0800 12 800, Lumi on 0800 99 533, Safeonweb, Child Focus and pointdecontact.be for illegal content, local police or 112 in an emergency. Everyone referred to here is an adult.
Why fakes have changed the nature of blackmail
Classic blackmail born out of a video chat followed a simple logic: the abuser pushed the person to expose themselves, recorded it, then threatened to share it. Our existing guide on homophobic entrapment and blackmail after a video chat describes this pattern and the reflexes to adopt.
Fakes turn the equation on its head. The abuser no longer needs your cooperation: they need your face and your name. A random video chat hands over the first in a matter of seconds, and the second is often deduced from everything else — a first name let slip, a city, a recognisable background, a username reused elsewhere. Consumer image-synthesis tools, widely documented since 2023 by bodies such as l'ANSSI (the French National Cybersecurity Agency) and by the work of the European Union Agency for Cybersecurity (ENISA) on information manipulation, can produce an image that looks convincing to the naked eye in a few minutes, on a phone.
Three practical consequences:
- Being careful about content is no longer enough. Showing nothing intimate does not protect you from a fabricated image. Many people believe they are safe because they "didn't do anything".
- The target of the blackmail is not the truth, it is the people around you. A relative who receives an image by message does not run a forensic analysis: they react emotionally.
- Fakes are fragile, legally and technically. A doctored image leaves traces, and the law now addresses it by name. That is your advantage — provided you do not give in during the first hour.
Spotting a fake without being an expert
You do not have to prove the forgery yourself, but knowing how to spot it changes your stance in the face of the threat. The most useful clues are often crude ones.
Visual giveaways
- The joins around the neck, ears and jaw. A difference in skin tone, an outline that is too sharp or too blurred, an ear partially erased.
- Elements that do not replicate well: glasses, piercings, a chain, a watch, a mole. A piece of jewellery that switches sides or vanishes from one image to the next.
- Teeth, hair, hands. These are still the most unstable areas for generators.
- Inconsistent lighting: a face lit from the front on a body lit from the side, or a missing cast shadow.
- Text inside the image: interface timestamps, platform names, banners. Invented interfaces often contain labels that do not exist or are misspelled.
Contextual giveaways
A screenshot supposedly taken from a video chat has to be consistent with the reality of the platform: resolution, framing, window position, interface language. A background that is not yours, a wall you have never had behind you, a camera angle that is impossible with your equipment — all of that can be demonstrated. That is also why knowing what your background says about you on video chat works both ways: your real background becomes proof of authenticity, or of forgery.
Metadata
An image that has gone through a generation or editing tool loses or alters its EXIF data. Some files now carry provenance markers (the C2PA / Content Credentials initiative, backed notably by Adobe and several camera manufacturers) and invisible watermarks. Do not try to read them by modifying the file: hand over the original, and let the investigators or a lawyer have it analysed.
The first forty minutes: what is really at stake
The blackmailer is banking on panic. Their message almost always includes a countdown, a modest sum to begin with, and a list of people close to you to show they have done their homework.
Do not pay. The reports collected by Cybermalveillance.gouv.fr on sextortion repeat it year after year: a first payment never buys silence, it proves the target pays.
Delete nothing. Not the account, not the conversation, not the messages. You would be destroying your own evidence.
Stop replying, but do not block straight away if that channel is your only source of evidence. Capture first, block afterwards.
Building the file, in order
- Full screenshots showing the profile, the timestamp, the URL and the entire thread — not cropped excerpts.
- An export of the conversation where the platform allows it, plus the list of accounts and usernames used.
- The original file you received (do not open it in an editor, do not recompress it, do not resend it through a service that degrades it).
- A dated log: date, time, channel, what was demanded, what was sent. A simple spiral notebook kept by hand does the job perfectly and stays readable even if your phone is seized or lost.
- A copy away from the device, on an encrypted USB stick kept somewhere other than next to the computer, or in a digital vault to which you alone hold the key.
The detailed method for preserving evidence is set out in our guide on reporting, blocking and keeping evidence after a bad exchange.
What French law has said since 2024
This is the point most blackmailers are unaware of, and the one you need to know.
| Act | Legal basis | Penalties |
|---|---|---|
| Sharing a doctored recording of someone's words or image without consent | Art. 226-8 of the code pénal (French Criminal Code) | 1 year's imprisonment and a €15,000 fine |
| Sharing a sexual deepfake without consent | Art. 226-8-1, created by the law of 21 May 2024 visant à sécuriser et réguler l'espace numérique (on securing and regulating the digital space, known as SREN) | 2 years and €60,000 (aggravated where shared online) |
| Blackmail | Art. 312-10 | 5 years and €75,000 |
| Extortion (where payment is obtained under threat) | Art. 312-1 | 7 years and €100,000 |
| Moral harassment, including online and by several people | Art. 222-33-2-2 | up to 3 years |
| Invasion of privacy | Art. 226-1 and 226-2 | 1 year and €45,000 |
Two decisive points:
- A homophobic motive is an aggravating circumstance under French law when the offence is committed because of the victim's sexual orientation, real or presumed (art. 132-77 of the code pénal / French Criminal Code). Say so explicitly when filing your complaint, and ask for it to be recorded.
- Article 226-8-1 does not require the content to be real. Quite the opposite: it targets precisely content that has been generated or manipulated. The fact that the image is fake does not weaken your complaint, it is its very basis.
In Belgium, the non-consensual sharing of sexual images is punished by article 417/11 of the code pénal (Belgian Criminal Code) — provisions introduced by the 2022 reform of sexual criminal law — blackmail by article 470, and discriminatory motives are taken into account by the anti-discrimination law of 10 May 2007, whose implementation is monitored by Unia.
Getting the content removed, fast
Takedown and the criminal complaint are two parallel tracks. Do not make one depend on the other.
Channels to activate the same day
- PHAROS (internet-signalement.gouv.fr) for illegal content online, in France.
- The host platform's reporting form: social networks, image hosts, messaging apps. Since the European Digital Services Act (DSA) came into force, large platforms must provide a traceable reporting mechanism and give reasons for their decision. Keep the acknowledgement of receipt.
- 3018, in France, for digital violence: the platform has fast-track reporting channels with the main networks and offers free support.
- De-indexing requests to search engines: Google and Bing process removal requests for non-consensual intimate images, including synthetic ones. The GDPR right to erasure (art. 17) applies, and the CNIL (French data protection authority) can be petitioned if you are refused. Our article on screenshots, re-identification and the right to be forgotten explains this process in detail.
- In Belgium: pointdecontact.be and Safeonweb.
Monitoring the spread without drowning in it
A reverse image search (Google Images, Bing Visual Search, TinEye) once or twice a week is enough to catch a reappearance. Set an alert on your name and your usernames. Give yourself a fixed slot and stick to it: checking twenty times a day adds nothing and takes a heavy toll.
Protecting the people around you before the abuser reaches them
This is the part technical guides forget, and the one that makes the difference.
If the abuser threatens to write to people close to you, you usually have a few hours' head start. Getting ahead of it with one or two trusted people removes blackmail's only lever. You do not have to come out in order to do so: one factual sentence is enough — "someone is trying to blackmail me with a faked image and may write to you; don't reply, forward everything to me". Plenty of people have defused blackmail in three messages.
If outing becomes unavoidable, the guide on limiting the spread and taking back control of the narrative explains how to regain the initiative, and the one on the aftermath of online homophobic abuse covers how to absorb the shock and who to talk to.
If your job is at risk, the article on what happens when it reaches the workplace sets out protection against discrimination based on sexual orientation, in both France and Belgium.
Reducing the raw material: what to do beforehand
You cannot eliminate the risk, you make it expensive. The goal is simple: that a capture of you should allow neither a convincing fake, nor an identification.
- Separate your face from your name. A username that exists nowhere else, a dedicated email address, no links to your social media. That is the foundation described in our discretion guide for people who are not out yet.
- Break up the visual markers of your room. A neutral background, a stretched sheet, a collapsible backdrop, a dark acoustic panel behind you: fewer identifiable details, and a setting a fake will struggle to reproduce.
- Keep the camera off by default. A sliding webcam cover remains the most reliable move on a laptop: it depends on no software at all.
- Reduce the number of available angles. Varying your lighting and position paradoxically gives a generator more material to work with. Stable lighting and a single framing limit the usable variants.
- Lock down the accounts that could serve as relays. A password manager and two-factor authentication stop an abuser from sharing the fake from your own accounts — a frequent scenario, covered in our article on hacked access. A physical FIDO2 security key closes that door for good.
- Know what your connection reveals, by reading our guide on what your IP address really gives away.
What if the fake circulates in a country that criminalises homosexuality?
The situation changes in nature. In several countries, a doctored image is not merely an invasion of your privacy: it becomes a piece of evidence used against you. Human Rights Watch documented, in its report "All This Terror Because of a Photo" (2023), how images and conversations extracted from phones have served as the basis for prosecutions and violence in the Middle East and North Africa. Amnesty International has reported waves of arrests in Tunisia in 2025 and the shrinking of online civic space in Uganda following the 2023 law.
In that context:
- never pass the file on to local acquaintances "to check";
- do not file a complaint locally without assessing the risk of it being turned against you — that is the central point of our article on Algeria, Tunisia and Morocco;
- if you are travelling, prepare your device before the border, as explained in the guide on travelling to a country that criminalises homosexuality;
- contact the relevant NGOs rather than the police: Amnesty International, Human Rights Watch, ILGA World, and, for French nationals abroad, the Centre de crise et de soutien du ministère de l'Europe et des Affaires étrangères (Crisis and Support Centre of the Ministry for Europe and Foreign Affairs).
Three sentences to keep handy
Blackmail is mostly won by refusing to enter the conversation the abuser has prepared.
"I am not paying. The content is fake and I have kept every one of your messages."
"Any distribution will be the subject of a criminal complaint, including for blackmail and for sharing a doctored image without consent."
Then: nothing more. No further replies, no negotiation, no explanations.
After that, you move to execution: screenshots, complaint, reports, two trusted people warned. You can also read our /safety page for general reflexes, and the video chat glossary if a technical term in a reporting form trips you up.
Key takeaways
- A fake does not require you to have done anything: being careful about content does not protect you from fabrication.
- Being fake is a legal asset, not a weakness: in France, article 226-8-1 of the code pénal (French Criminal Code) explicitly targets sexual deepfakes shared without consent, since the SREN law of 21 May 2024, and a homophobic motive increases the penalties.
- Pay nothing, delete nothing, capture everything, and pursue takedown and complaint in parallel.
- Warning two trusted people disarms blackmail better than any technical measure.
- Reduce the raw material: separate face from identity, neutralise your background, lock down your accounts.
You have nothing to reproach yourself for. The person fabricating an image to silence you is the one committing several offences — and in practice, they are easier to document than he thinks.
