Content warning: this article discusses image-based sexual abuse, non-consensual intimate images and AI-generated sexualised imagery.

AI debates often focus on copyright, automation, productivity and business efficiency. Those issues matter. But image-based sexual abuse shows the sharpest version of the consent problem.

Image-based sexual abuse happens when someone creates, takes, shares or threatens to share intimate images without consent. It can include real images, edited images and AI-generated fake images that appear real.

Police.uk describes non-consensual intimate image abuse as including someone creating, or asking someone to create, a fake intimate image or video that appears real without permission. It describes this as a serious sexual offence.

That matters because AI has changed the practical risk. A person no longer needs specialist technical skill to create a convincing fake. A normal photograph can become the starting point for harassment, humiliation, coercion or abuse.

The victim may never have created or shared any intimate image. The harm still targets their identity, body and dignity.

Why language matters

People still use the phrase "revenge porn". The term remains common because the main UK support service carries that name, and many people recognise it. But the phrase has serious problems.

·        It suggests revenge, when the motive may include coercion, entitlement, misogyny, humiliation, control, extortion or cruelty.

·        It uses the word "porn", which suggests consensual sexual content.

·        It can shift attention toward the victim's behaviour rather than the perpetrator's abuse.

The better terms are image-based sexual abuse, intimate image abuse or non-consensual intimate image abuse. Those terms describe the conduct more accurately. They put the absence of consent at the centre.

Credit to campaigners

Jess Davies deserves credit for her campaign work on this issue. She has helped bring public attention to digital sexual abuse, deepfake abuse and the victim-blaming that often surrounds intimate image abuse. Campaigners like her make the issue harder to dismiss as a private embarrassment or a niche online problem.

It is not a private embarrassment. It is abuse.

The Grok controversy

The controversy around Grok and X showed how quickly AI image tools can become harmful when platforms fail to design sufficient safeguards. Reports described users generating sexualised or manipulated images of real people, including concerns involving children. AP reported that UK authorities moved to criminalise creating or requesting deepfake sexual images without consent and that Ofcom had launched an investigation into whether X breached UK laws in connection with Grok-generated images.

The important point is not only whether one platform broke rules. The wider question is whether AI providers, platforms and regulators can keep pace with abuse patterns.

Law has moved, but harm moves faster

UK law has developed. Guidance now recognises fake intimate images and deepfakes within non-consensual intimate image abuse. But victims still face practical barriers.

·        They may need to find where an image appears.

·        They may need to collect evidence without spreading it further.

·        They may need to report to platforms.

·        They may need to report to police.

·        They may need emotional support.

·        They may fear that removal from one site will not stop re-uploading elsewhere.

That process can place too much burden on the person already harmed.

What responsible AI governance should ask

Any platform or organisation that uses AI image tools should ask direct questions:

·        Can the tool create sexualised images of real people?

·        Can it alter a normal image into an intimate or suggestive image?

·        Can it create images that resemble children or young people?

·        Does the tool block harmful requests before output appears?

·        Does it detect attempts to bypass safeguards?

·        Can victims report content quickly?

·        Can platforms remove content quickly?

·        Can law enforcement get evidence when needed?

·        Does the provider treat this as predictable misuse rather than an edge case?

A balanced position

AI image tools can have legitimate uses. They can support education, accessibility, design, storytelling and creativity. The issue is not image generation itself.

The issue is consent.

A tool that can manipulate real people's likenesses needs safety by design. It should not rely on victims discovering the harm after publication. It should not place the main burden on the person targeted. It should not hide behind the idea that misuse came from users alone when the system made that misuse easy.

This article sits within a wider series about AI, ownership and responsibility. Consent links every topic.

·        Artists need consent over their work.

·        Organisations need control over their data.

·        People need consent over their likeness, body and sexual identity.

When AI ignores consent, the result can move from legal uncertainty to real personal harm.

Support links

Revenge Porn Helpline: https://revengepornhelpline.org.uk/

Police.uk guidance on non-consensual intimate image abuse: https://www.police.uk/advice/advice-and-information/online-safety/online-safety/non-consensual-intimate-image-abuse-revenge-porn/

StopNCII.org: https://stopncii.org/

Samaritans: 116 123 or https://www.samaritans.org/how-we-can-help/contact-samaritan/