AI‑Generated Indecent Images: How the Law Is Changing and What It Means for You

How the Law Handles AI-Generated Indecent Images vs Real Images 

From a defence perspective, a crucial question is: Are AI-generated images treated the same as “real” indecent images of children, even at the sentencing stage?  

Yes – largely they are treated the same, because the law doesn’t distinguish the harm by the image’s origin. The content and context (what’s depicted, how widely shared, the offender’s intent) tend to influence the outcome, not whether an image was computer-generated or not. 

Child deepfake images 

In court, AI-made images (pseudo-photos) are classified into the same Category A/B/C scale as actual child photographs, based on the severity of the content. The Sentencing Council guidelines apply equally to pseudo images as to real images. For example, an AI image depicting a child in penetrative sexual activity is Category A (most serious), just like a real photo would be. A person convicted of making or distributing such images faces up to 10 years’ imprisonment (the statutory maximum under PCA 1978), and likely a multi-year sentence depending on volume and categories (often those dealing in AI images also have numerous images). In fact, recent cases confirm courts’ hard line:  

In R v Jaycock (2024), a man created indecent images by AI face-swapping children’s faces onto adult pornography. The Court of Appeal held these were correctly treated as “production” offences (i.e. making new images) with a high sentence range. He ultimately received 3 years 10 months in prison for a small number of AI images, illustrating that even a few generated images can yield years of custody. The judges noted such deliberate digital creation is a serious crime, exploiting children’s likenesses for sexual gratification.  

Another case in 2024 at Bolton Crown Court saw a defendant use AI to turn everyday pictures of real children into child abuse images, which he then sold to other offenders. He was caught and handed an 18-year sentence (plus 6 more on extended licence), reflecting the very high seriousness when such material is produced and distributed at scale.  

These outcomes underscore: from the law’s view, AI images of child abuse are child abuse images. A police spokesperson put it bluntly after a 2025 sentencing: “This depraved act involving the use of AI is treated just as severely as using actual photographs”.  

Adult deepfakes images 

Historically, if you made a deepfake of an adult but didn’t share it, there was no specific offence, though other legislation like harassment or obscene publications might have been considered. But as soon as you sent or posted it, it became illegal under image abuse laws (the 2015 “revenge porn” law and now updated offences in 2023).  

With the new deepfake creation offence, creating an explicit fake of someone without consent, even if it stays on your device, can lead to a prosecution. That offence, once active, carries an unlimited fine on conviction.  Also, in any case with a named victim (like a deepfake of an actual person), expect restraining orders or other protective orders in addition to criminal punishment.  

Will I go to prison for AI generated indecent images? 

The sentencing guidelines for indecent images of children are well-established (with factors like the number of images, categories, distribution, etc., determining length of sentence). There is no separate guideline for “AI images” – they are encompassed by the existing framework. If anything, defence lawyers may argue that because no real child was directly harmed, a sentence could lean towards the lower end of the range. However, courts have so far not shown much leniency on that basis. The presence of AI might be seen as an aggravating factor if it indicates advanced planning or potential distribution at scale.  

In sum, the law’s approach is to equate AI indecent images with real ones, from criminal liability to sentencing. For someone accused, that means you should not expect any “technicality” to lessen the seriousness: an AI category A image will be treated as category A offences. 

Can the Police Detect AI-Generated Images? 

Yes, and they are actively trying to identify and stop AI-generated indecent images. The truth is, many offenders are caught not because the police spot an image is AI, but because of digital traces or tips (like an IP address, an undercover operation, or a referral from the IWF). Once an investigation starts, forensic analysis of devices often reveals the use of AI tools or the presence of certain image files, even if the images look “real” to the naked eye.  

That said, AI can indeed make detection harder in some ways: 

• Police officers have admitted that some AI images are so lifelike, it’s difficult to visually tell them apart from real photos. If images are discovered on a suspect’s device, forensic analysts may use additional clues (like metadata, known hashes of real images, or evidence of AI software usage) to determine their origin. However, as far as the criminal charge goes, it doesn’t particularly matter if they confirm the image is AI or not – it’s illegal either way, and often multiple images (some real, some AI) are found together.  
• The Internet Watch Foundation (IWF) and other organisations are now specifically tracking AI-generated CSAM online. They have started flagging content that appears synthetic and working with tech companies. The IWF’s recent data show they categorized hundreds of AI images (and often can recognise patterns). This means that if you share AI abuse material online, you stand a good chance of it being detected and traced back via known websites or networks.  
• Generative AI tools themselves leave some fingerprints. Large image-generation models might have subtle artifacts or known quirks, and researchers are developing detection algorithms to spot an AI-generated image. The UK government, under the Online Safety framework, is pushing platforms to use such technologies and expedite removal of both real and deepfake images.  

So, while an AI image might fool the human eye, police and experts have many ways to uncover the crime. One scenario is that someone could be arrested for unrelated online activity (say, a message or tip-off), leading to device seizures that reveal AI child abuse images. Devices likely contain evidence of the generation process (like installed AI software, prompt history, cached files), making it even clearer. In summary: you should assume AI images are discoverable and prosecutable – there’s no “safe hiding” behind technology. 

For those worried about inadvertently stumbling on such content: If you ever encounter what looks like an AI indecent image (for example, forwarded in a group chat), do not forward it or save it. The best action is to delete it and consider reporting to authorities (IWF has a hotline). Possession itself is an offence – but authorities focus on those deliberately generating or collecting such material rather than incidental recipients. If in doubt, seek legal advice immediately on how to protect yourself. 

What’s Changing? 

We’ve touched on new laws above, but let’s clearly outline recent and upcoming legal changes specifically relevant to AI-generated sexual images: 

• Online Safety Act 2023 – This landmark Act (which received Royal Assent in October 2023) did many things, but the crucial part was creating new intimate image abuse offences and updating existing law. It inserted new sections 66B, 66C, 66D into the Sexual Offences Act 2003, covering sharing intimate images without consent. These offences explicitly include images that “appear to show” a person in an intimate situation. That phrase was chosen to ensure that deepfake images (which appear to show someone naked or in a sexual act) are covered, even if they are fake. The Act also criminalised threatening to share intimate images (closing a gap in the old law).  
• New offence of creating deepfakes (2026) – As noted, by early 2025 the government announced a standalone offence to criminalise intentionally making a sexual deepfake without the depicted person’s consent. This targets situations where someone might create a humiliating nude image of someone just to harass or gratify themselves, without necessarily disseminating it. This was introduced in February 2026. It means if you privately create a deepfake of Person X, you could be charged even if you never hit “send”. And if you do share it, then you could be charged with both the creation offence and the sharing offence from the Online Safety Act.  
• Focus on AI tools  – The offence isn’t limited to the hands-on creator. The new 2026 legislation also criminalises requesting or causing someone else to create a non-consensual explicit deepfake under section 66F of the Sexual Offences Act 2003. In other words, if Person A asks or pays Person B to make a fake sexual image of Person C, both A (who instigated it) and B (who did the editing) could be guilty of a criminal offence under the law.   This ensures those who procure deepfakes (for example, paying a website to produce a fake pornographic video of an ex-partner) cannot escape liability by saying “I didn’t do it myself.” This signals the intense direction of travel: not just punishing images, but also those enabling their creation.  
• Online Safety regulatory changes – Under the Online Safety Act, new rules mean platforms and tech companies now have legal duties to swiftly remove indecent images (including deepfakes). For instance, from late 2024, “sharing intimate images” was made a priority offence that platforms must address. This regulatory environment means it may become harder for such content to circulate openly online (forcing it further underground, where law enforcement focus increases).  

What does this mean for individuals?  

Essentially, the net is tightening. By 2026, nearly every corner of “image-based sexual abuse” – whether real or AI, child or adult – is  explicitly illegal. Defence solicitors must stay up to date because cases may involve brand-new laws or charges. If you are investigated, you need a legal team who understands these developments and can navigate the intricacies (for example, challenging evidence of who created an image, or whether the prosecution can prove intent if the law requires it).  

If You’re Under Investigation for AI-Generated Images 

Facing a criminal investigation for indecent images or deepfake offences is incredibly daunting. If AI-generated images are involved, the case can be complex, but a strong defence is possible. Here’s what you need to know and do: 

1. Seek Specialist Legal Advice Immediately: As soon as you suspect you’re under investigation (police contact, device seizure, etc.), contact a specialist solicitor experienced in indecent images and cyber offences. These cases involve technical evidence (forensic computer analysis, possibly AI tools usage logs) and rapidly evolving law, so a specialist is essential. Do not attempt to “explain it away” to police without a lawyer present. Even seemingly innocuous statements (like “I was just experimenting with an AI model”) can seriously harm your defence later. 
2. Understand the Scope of Potential Charges: Based on current law, you could face multiple charges: for instance, one charge for “making indecent images” (if any AI child images are found) and if adult deepfakes of specific people are found, you could face charges under the new intimate image abuse offences for sharing (if applicable), or eventually the creation offence. Each has different elements (for example, the deepfake creation offence requires proof you lacked the person’s consent and did it for sexual gratification or to cause distress). A seasoned defence lawyer will parse these details and possibly argue that some images don’t meet a certain offence definition.  
3. Preserve Evidence and Context: Offenders with AI images sometimes also have legal but contextually relevant materials, for example, artificial “nudifying” apps, image editing software, or benign images that were used as bases. Don’t delete or destroy anything after the fact (that can be a separate offence and make you look worse). Instead, let your solicitor know of anything that could help explain or contextualise the allegations. For example, if an AI model inadvertently produced something you didn’t intend, or if you were engaged in research, these could be important points to raise (though they may not excuse possession, they could influence decisions). 
4. Expert Involvement: With complex computer evidence, defence teams often engage independent computer forensic experts who can review how images were created or stored, and AI experts if necessary. They might check if image detection technology could mistakenly flag something(especially important if you genuinely didn’t know something was on your device). In deepfake cases, we might need experts to discuss face-mapping techniques or authenticity, especially if there’s a dispute whether an image is a deepfake or genuine. This can raise reasonable doubt or affect the prosecution’s narrative. 
5. Pre-Charge Representations: At the pre-charge stage, a proactive defence solicitor can make representations to the police or CPS. For example, your lawyer could highlight any lack of evidence of distribution or argue that images found might not meet the legal definition of “indecent” (in borderline cases) or that the case lacked a reasonable prospect of onviction.   In other cases the evidence may be strong but your legal team could argue that it is not in the public to prosecute and that an out of court disposal such as a caution may be suitable. The aim is to influence the charging decision, possibly persuading them to proceed with lesser charges or none at all if the case is weak. 
6. Prepare for Ancillary Consequences: If charged, besides the core penalties, Sexual Offences Prevention Orders or Sexual Harm Prevention Orders often accompany indecent image convictions (restricting internet use, etc.), and Sex Offenders’ Register notification is typically mandatory.  

Ultimately, every case is unique. A robust defence will challenge the prosecution to prove every element, from proving the images are indecent by legal standards, to proving you made or possessed them knowingly. The evolving nature of AI may offer novel defence angles – perhaps questioningwhether you had the intent to create illegal images if an AI unexpectedly generated them. A top defence team stays at the cutting edge of such arguments.