Sexual deepfakes: when AI becomes a tool of violence

An ordinary photo posted on social media can be repurposed to fabricate fake intimate content. The image is artificial, but the consequences are entirely real. Sexual deepfakes raise a new question about consent in the age of artificial intelligence.

What is a sexual deepfake?

A deepfake is visual or audio content manipulated or generated using algorithmic techniques, so as to give the impression that a person did, said or showed something that never actually happened.

In its sexual form, the technology can be used to fabricate an intimate image or video of someone without their agreement. So this is not merely a problem of “false information”. The harm strikes directly at intimacy, reputation, dignity, and control over one’s own image.

In France, distributing this content is punishable

The French Criminal Code explicitly covers sexual montages and content created or generated from a person’s image or words without their consent.

Article 226-8-1 provides in particular that bringing such content to the knowledge of the public or of a third party is punishable by two years’ imprisonment and a fine of 60,000 euros. Where it is published through an online public communication service, the penalties can reach three years’ imprisonment and a fine of 75,000 euros.

The essential point is simple: the fact that an image is “fake” does not make distributing it harmless.

Why this deserves real prevention work

Generative tools are making synthetic content easier and easier to create. Recent research points out that responses focused only on detecting the “fake” miss part of the problem: even when everyone knows an image is artificial, the person depicted can still face humiliation, harassment or damage to their reputation.

Digital prevention must therefore take on board something essential: consent also applies to the use of our image.

What to do if you come across such content

Avoid resharing it, even to call it out: every new circulation can amplify the harm. Instead, keep material that documents the content’s existence, such as the page address, the account involved, the date and any context. Use the platform’s reporting tools and, depending on the situation, get support to take the appropriate steps.

For friends and family, one principle matters above all: believe, support, and do not lay blame. Having posted a photo of yourself gives nobody the right to sexualise it or repurpose it without consent.

Consent has to become digital too

For a long time, prevention work around consent mainly concerned physical interactions. AI now forces us to widen that thinking: consenting to publish a photo does not mean consenting to it being altered, sexualised, or circulated in another context.

This is a new challenge for digital education, and also a very concrete part of the fight against sexist and sexual violence.