Stay Informed
Join our newsletter to receive updates about our mission, impact stories, and ways you can help make a difference.
ChatGpt and child abuse
Written by:
Lena Kyony
•
The clock is ticking. Every day we wait to address AI-generated child sexual abuse material is another day predators get ahead of law enforcement.
OpenAI recently announced that ChatGPT will permit erotic content for age-verified adults. The company has framed this decision as a way of respecting adult autonomy. But this decision, opens a floodgate that could accelerate one of the most urgent threats facing children today: the explosion of AI-generated child sexual abuse material (AI-CSAM).
The numbers are already terrifying. And they're getting worse by the day.
The Crisis Is Already Here!
This isn't a hypothetical future threat. Here are the facts and figures! According to the internet Watch Foundation (IWF) July 2024 report, over 3,500 new AI-generated child sexual abuse images appeared on a single dark web forum between 2023 and 2024. The most concerning aspect of this discovery is that the images are becoming more severe, with more depicting the most extreme Category A abuse.
In 2024 Reports of AI-generated CSAM surged 380% from 51 confirmed reports in 2023 to 245 in 2024. Each report can contain thousands of individual images. The National Center for Missing & Exploited Children received 67,000 reports of AI-generated CSAM in all of 2024 then received 485,000 reports in just the first half of 2025. That's a 624% increase in six months.
This is a major concern as Law enforcement can no longer tell what's real. In 193 of 245 reports investigated, AI-generated images were so realistic they had to be treated as if they depicted actual children being raped. The problem is that it’s difficult to justify prosecuting those who create these images, since they don’t involve real people but are instead artificial depictions.
We are losing this fight. And loosening restrictions on sexual AI content will make it exponentially worse.
The Human Cost of Delay
These aren't just statistics. Real predators are already weaponizing AI to abuse children:
Hugh Nelson (UK, 2024)
A British man was sentenced to over 20 years in prison for using AI face-transformation software to convert normal photographs of real children into graphic sexual abuse imagery. He then sold albums of these images in online chat rooms on the open web. Nelson targeted real children turning innocent photos into nightmare fuel that will haunt those victims forever.
Steven Andereg (Wisconsin, 2025)
The first federal prosecution for creating CSAM entirely through AI. Anderegg faces charges for producing images that never depicted a real child yet prosecutors argue (correctly) that such material still harms children by normalizing abuse and fuelling demand for exploitation.
Europol Operation (February 2025)
Dozens arrested across Europe after a Danish man allegedly ran a paid platform distributing fully AI-generated child sexual abuse material. 273 suspects identified. Users paid to access content showing "children being abused." The platform's administrator allegedly produced the content using AI tools.
California's Eight Lost Cases
Between December 2024 and September 2025, prosecutors in Ventura County, California, had to drop eight cases involving AI-generated CSAM because state law required proving the imagery depicted a real child. Eight predators walked free because the law couldn't keep up with technology.
Kaylin Hayman (Disney Channel Actress)
A 17-year-old actress from Disney's "Just Roll with It" became a victim of AI-generated "deepfake" sexual imagery. "I felt like a part of me had been taken away," she said. "Even though I was not physically violated." Her advocacy helped pass California's updated law — but millions of children remain unprotected.
Why Allowing Adult Erotica Makes the Child Abuse Problem Worse
When AI systems are trained and optimized to produce sexual content for adults, five devastating side effects follow:
1. The Supply Explodes
Models designed to generate adult sexual content can be manipulated through clever prompts, jailbreaks, or outright abuse to produce sexualized images of minors. Even with filters, bad actors find workarounds. The technical capability to create erotica for adults is the same capability that creates AI-CSAM.
2. Synthetic Abuse Is Real Abuse
A dangerous myth persists: "It's not real, so it's not harmful." This is false.
AI-CSAM normalizes the sexual abuse of children
It can be used as a grooming tool to desensitize real children
It re-traumatizes survivors who see that society is creating new ways to depict their abuse
It trains and rehearses predators, lowering their inhibitions
Studies show it can stimulate offending behavior
Every child-safety organization on Earth agrees: synthetic child sexual abuse material is abuse material, full stop.
3. Age Verification Won't Save Us
OpenAI says adult content will be "age-gated," but:
Age verification systems can be spoofed or bypassed
They create new privacy and data security risks
They can't stop verified adults from creating and distributing AI-CSAM
Even Disney Channel actresses get deepfaked verification doesn't protect victims
4. Detection Systems Are Already Overwhelmed
Traditional child-protection tools rely on "hash matching" comparing new images to databases of known abuse. AI-generated content bypasses these systems entirely because every image is new and unique.
Law enforcement is drowning:
98% of AI-generated CSAM depicts girls amplifying violence against women and girls
Investigators waste time trying to identify and rescue children who don't exist
The sheer volume threatens to collapse detection and reporting systems
"We're playing catch-up to a technology that is moving far faster than we are," says California DA Erik Nasarenko.
5. Grooming and Coercion Get an AI Upgrade
Sexual chatbots can be weaponized as grooming tools:
Predators can role-play abusive scenarios and refine their tactics
AI can generate personalized sexual content featuring a specific child's likeness
Chatbots can be used to normalize inappropriate relationships and boundary violations
The line between fantasy and action blurs when AI makes abuse "feel" accessible
What Must Happen Right Now
Every day we delay, predators generate thousands more images. Here's what needs to happen immediately:
1. Legal Clarity: All AI-CSAM Is CSAM
an AI-CSAM Is child sexual abuse material generated by AI
Every jurisdiction must unambiguously criminalize AI-generated child sexual abuse material whether it depicts a real child or not, whether it's photorealistic or cartoon-like. The law must catch up to the technology now. As of late 2025, many U.S. states still have legal gaps. Children cannot wait for lawmakers to feel comfortable.
2. Strong Age Verification
Age verification should be robust, privacy-respecting, and independently audited. But it's only a speed bump and not a permanent solution. Verified adults can still commit crimes. We need multiple layers of protection, not a single gate.
3. Mandatory Watermarking and Provenance
Every AI-generated image should carry tamper-resistant metadata or forensic watermarks. This won't stop abuse, but it will make it harder to deny and easier to trace. Investigators need tools to distinguish real photos from AI creations and to track who made what.
4. Next-Generation Detection, Funded Now
Traditional hash-matching is useless against novel AI content. We need machine learning systems that can detect patterns, styles, and content in AI-generated imagery. This technology exists. It needs funding, deployment, and mandatory use by platforms.
5. Mandatory Reporting and Transparency
Platforms must report suspected CSAM and AI-CSAM to national hotlines (like the National Center for Missing & Exploited Children in the U.S.). They must share threat intelligence with law enforcement. And they must publish transparency reports showing how much abuse they detected, how much they missed, and what they did about it.
6. Fund Victim Support and Civil Society
Survivors need resources. Advocacy groups need funding. Researchers need access. We cannot leave child protection to tech companies and law enforcement alone. Civil society must have a seat at the table and the resources to fight back.
The Bottom Line: This Is an Emergency
Treating adults like adults is a reasonable principle but not when the tool in question is a machine that can generate infinite child sexual abuse material on demand.
OpenAI and every other AI company must understand: loosening restrictions on sexual content without ironclad safeguards is reckless. We are at a fork in the road. We can act now with clear laws, strong safeguards, mandatory detection, and real consequences or we can watch as AI becomes the greatest accelerant of child sexual abuse in history.
There is no middle ground. The time to act is now.
For more information, see:
Internet Watch Foundation: IWF AI-CSAM Research
National Center for Missing & Exploited Children: NCMEC
ECPAT International: ECPAT
U.S. Department of Justice Child Exploitation Section: DOJ
References
News & Policy Sources
Reuters. (2025, October 14). OpenAI to allow mature content on ChatGPT for adult verified users starting December 2025. Retrieved from https://www.reuters.com/business/openai-allow-mature-content-chatgpt-adult-verified-users-starting-december-2025-10-14
The Verge. (2025, October 15). Sam Altman says ChatGPT will soon sext with verified adults. Retrieved from https://www.theverge.com/news/799312/openai-chatgpt-erotica-sam-altman-verified-adults
Ars Technica. (2025, February). ChatGPT can now write erotica as OpenAI eases up on AI paternalism. Retrieved from https://arstechnica.com/ai/2025/02/chatgpt-can-now-write-erotica-as-openai-eases-up-on-ai-paternalism
National Center for Missing & Exploited Children (NCMEC). (2025). CyberTipline Data Report, First Half 2025. Alexandria, VA: NCMEC.
https://www.missingkids.org/ncmec-data-reports
Internet Watch Foundation (IWF). (2025). Annual Report 2024–2025: AI-Generated Child Sexual Abuse Imagery. Cambridge, UK: IWF.
https://www.iwf.org.uk/report/2024
Europol. (2025, February). Press release: Major European operation targets creators and distributors of AI-generated child sexual abuse material. The Hague: Europol.
https://www.europol.europa.eu/media-press/newsroom
BBC News. (2024, September 9). Man jailed for using AI to create child abuse images. London: BBC.
https://www.bbc.com/news/technology
U.S. Department of Justice. (2025, March 12). Wisconsin man indicted for creating child sexual abuse material using artificial intelligence. Washington, DC: DOJ Press Release.
https://www.justice.gov/opa/pr
Ventura County District Attorney’s Office. (2025, September). AI-generated child sexual abuse cases dismissed due to legal gaps. Ventura, CA: Press Statement.
Los Angeles Times. (2025, March 3). Disney Channel actress Kaylin Hayman speaks out on deepfake exploitation, calls for stronger AI laws. Los Angeles: LA Times.
Thorn. (2024). The Rise of AI-Generated Child Sexual Abuse Material and the Urgent Need for Detection Standards. Los Angeles: Thorn.
https://www.thorn.org/blog/ai-csam-detection-standards
WeProtect Global Alliance. (2024). Artificial Intelligence and Child Sexual Abuse Material: Emerging Risks and Policy Responses. London: WeProtect Foundation.
https://www.weprotect.org/library
ECPAT International. (2025). AI and Online Child Sexual Exploitation: Global Threat Assessment 2025. Bangkok: ECPAT.
https://www.ecpat.org/resources
United Nations Office on Drugs and Crime (UNODC). (2024). Global Report on Online Child Sexual Exploitation 2024. Vienna: UN Publications.
California District Attorneys Association. (2025, August). Statement on AI-Generated Child Sexual Abuse Material and Legislative Reform Needs. Sacramento, CA: CDAA.
Nasarenko, E. (2025, September). Interview with ABC7 Los Angeles on AI-CSAM and prosecutorial challenges. Retrieved from https://abc7.com
