AI-Generated Child Sexual Abuse Material in Canada
AI-generated CSAM overwhelms existing detection systems, complicates criminal prosecution by blurring the line between real and synthetic imagery, and creates new vectors for child exploitation. Canada's Criminal Code provisions on CSAM need to be tested and potentially updated for the generative AI era.
Narrative
The proliferation of generative AI image models has created a new and rapidly growing threat to child safety: the ability to generate photorealistic child sexual abuse material (CSAM) using text-to-image AI tools. The Canadian Centre for Child Protection (C3P), which operates Cybertip.ca and Project Arachnid, has identified AI-generated CSAM as an escalating concern, with reports of synthetic abuse imagery increasing significantly from 2023 onward.
Open-source image generation models can be fine-tuned or prompted to produce exploitative imagery of children. Unlike traditional CSAM, which documents actual abuse, AI-generated material can be produced at scale without requiring access to a victim — but it normalizes the sexualization of children, can be used to groom real victims, and overwhelms the detection infrastructure that organizations like C3P have built over decades. Hash-based detection systems like PhotoDNA, designed to match known CSAM images, cannot identify novel AI-generated content.
Canadian law addresses CSAM through Criminal Code provisions that cover visual representations depicting minors in sexual activity, which legal experts generally interpret as covering synthetic material. Prosecution of AI-generated CSAM cases is still in early stages — Steven Larouche of Sherbrooke, Quebec was sentenced in 2023 to over three years for creating deepfake child pornography, in what the presiding judge described as the first such case in Canada. The sheer volume of synthetic material threatens to divert law enforcement resources from cases involving real child victims, and distinguishing AI-generated from real imagery is increasingly difficult.
Canadian law enforcement, including the RCMP’s National Child Exploitation Crime Centre, and child protection organizations are calling for coordinated action: stronger model-level safeguards from AI developers, updated legal frameworks, new detection technologies, and international cooperation to address a transnational problem that accessible generative AI tools make worse.
Harms
Generative AI tools enabled the creation of photorealistic child sexual abuse material at scale, normalizing the sexualization of children, providing new vectors for grooming real victims, and overwhelming hash-based detection systems like PhotoDNA.
AI-generated CSAM blurs the line between real and synthetic abuse imagery, complicating criminal prosecution and threatening to divert law enforcement resources from cases involving real child victims.
Children depicted in or targeted through AI-generated exploitative material face psychological harm, including through the use of such material for grooming.
Affected Populations
- children
- law enforcement agencies
- child protection organizations
- online platforms
Entities Involved
Operates Cybertip.ca and Project Arachnid; identified AI-generated CSAM as an escalating concern and called for coordinated action including stronger model-level safeguards and updated legal frameworks
RCMP's National Child Exploitation Crime Centre is involved in investigating AI-generated CSAM cases and has called for coordinated law enforcement action
Responses & Outcomes
Issued public warning about AI-generated deepfakes of children, urging parents to be aware of the threat and calling for stronger protections
AI System Context
Generative AI image models, including open-source diffusion models, being used to create photorealistic child sexual abuse material. These tools can generate synthetic CSAM from text prompts or by modifying existing images, creating material that is difficult to distinguish from real abuse imagery.
Preventive Measures
- Clarify Criminal Code coverage of AI-generated CSAM to ensure synthetic material is unambiguously addressed
- Require generative AI model developers to implement safeguards against CSAM generation, including content classifiers and training data audits to remove exploitative material
- Fund the Canadian Centre for Child Protection and law enforcement to develop detection tools capable of identifying AI-generated CSAM alongside traditional hash-matching systems
- Mandate that platforms operating in Canada deploy both perceptual hashing and AI-based detection for synthetic CSAM, with reporting obligations to the National Child Exploitation Crime Centre
- Support international coordination through the Five Eyes and other frameworks to establish shared standards for AI model safety that prevent CSAM generation capabilities
Related Records
- calgary-teen-ai-csam-charges related
- Calgary Teen Charged with Creating AI-Generated Child Sexual Abuse Material from Classmates' Photos related
- AI-Generated Child Sexual Abuse Material in Canada related
Taxonomy
Sources
- Police and child protection agency say parents need to know about sexually explicit AI deepfakes
- National Strategy for the Protection of Children from Sexual Exploitation on the Internet
- Amid rise in AI deepfakes, experts urge school curriculum updates for online behaviour
AIID: Incident #604
Changelog
| Version | Date | Change |
|---|---|---|
| v1 | Mar 7, 2026 | Initial publication |