Deepfake Detection
for Investigating Sexual Crimes
As much as 98% of all deepfakes created globally are sexual in nature, with all content on the five most popular websites hosting pornographic deepfakes targeting women and girls. Almost all of this synthetic or AI-manipulated material is non-consensual intimate imagery (NCII) of real people – otherwise known as ‘deepnudes’.
In deepfake-driven sexual crimes, female celebrities are the most targeted group in terms of volume due to the ready availability of source material. But the problem also exists on a national and local level, with deepfake creation groups springing up in schools and universities to target the local population.
Many commercial generative AI tools can be used to create sexual imagery, but other generative models and services are also available, including bots on Discord and Telegram. Some sexual deepfake creation Telegram chat rooms have 200k subscribers, and Telegram bots can complete deepnudification requests autonomously and on demand.
Sexual Harassment
The technology for creating non-consensual intimate imagery (NCII), or deepnudes, is commodified, meaning synthetic or AI-manipulated video and image content can be created with ease. Original photos, pictures from social media, and other publicly available imagery all serve as source material for generating sexual deepfakes. Voices can also be convincingly cloned from just a few seconds of source audio.
Sextortion and coercion
Image-based sexual abuse (IBSA), or deepfakes based on real people can be used to extort, blackmail and coerce victims, including ‘revenge porn’ scenarios and pressuring victims into sending real compromising photos.
Reports of deepfake-related sex crimes in South Korea more than tripled in 2024, with the Digital Sex Crime Victim Support Centre receiving 1,384 reports of manipulated images used in cases in 2024, compared to 423 reports in 2023.
Deepfake CSAM and Indecent Images of Children (IIOC)
In 2025, the Internet Watch Foundation (IWF) assessed 8,029 AI-generated images and videos depicting child sexual abuse. Among them, 3,443 were videos – a staggering 26,385% increase on the previous year, when just 13 such videos were recorded. 65% of videos were classified Category A.
Deepfake sexual crime challenges
Deepfake-driven sexual crimes, including IBSA, NCII, and IIOC are a significant challenge for law enforcement due to the massive scale of the problem. As much as 99% of all deepfakes created globally are sexual in nature, and the vast majority of these are non-consensual images or videos of real women, and of children.
Many of the commercial and off-the-shelf generative AI tools lack guardrails to prevent the creation of such content, and digital watermarking efforts do not solve the problem.
Highly realistic synthetic sexual content can be generated from authentic videos and images of real people found online, while a clone of someone’s voice can be produced from just a few seconds of recorded audio.
The Crime-as-a-Service (CaaS) phenomenon, which automates and scales many aspects of criminal activity, such as Discord and Telegram bots capable of generating synthetic sexual content on demand, lowers the barrier to entry for newcomers, with pre-packaged tools and services available on criminal forums and the dark web.
Forensic-grade evidence analysis for investigating deepfake sex crimes
Scalable, automated and multilayer analysis with court-ready reports
Every result includes explainability, provenance metadata, and visual evidence supporting the conclusion. Reports are automatically generated in 30+ languages, enabling cross-border collaboration and ensuring admissibility in court.
Get actionable results in seconds, not hours
With standardized reporting, multilingual support, and strict chain-of-custody controls, Sensity AI transforms media authentication into actionable, certifiable intelligence. Investigators can export structured evidence packages for prosecutors, international partners, or judicial authorities.
Stay on top of the caseload
With millions of deepfakes in circulation, digital forensic investigators are under increasing pressure to work through large case loads. Sensity AI integrates synthetic media and deepfake detection at the earliest stage of evidence ingestion, so every extracted media file can be automatically screened for signs of manipulation before it enters the investigative record, speeding up time to evidence.
Cutting edge digital forensics
Specialist digital forensic analysis tools have always existed, but they are not equipped for detecting sophisticated AI-generated falsifications at scale. Sensity AI analyzes video, images, and audio within a single interface, automatically detecting audio in video, and provides relevant and contextual analysis within the same dashboard in real-time.
