The Edmonton Police Service has stepped into a complex ethical debate by using an AI-generated facial image to try and identify a woman whose remains were found downtown last year. The release of the image by the Missing Persons Unit marks a significant shift in how Canadian law enforcement is willing to use synthetic media when traditional forensic methods hit a wall. The woman was discovered near Bellamy Hill in December 2024, but after months without a match, investigators turned to the digital forensics section to create a lifelike approximation.
This specific unit typically handles facial imagery and related technologies, a detail that draws a direct line between this new generative approach and the service’s history with facial recognition tools. While the police have not stated that the AI image was run through an automated recognition system, the use of synthetic imagery tied to a real person creates a functional overlap between forensics, biometrics, and generative AI. It forces the public to grapple with the difference between a forensic sketch, which is understood to be an artistic interpretation, and a photorealistic AI generation that claims to represent a real human being.
An ethicist interviewed about the case described the approach as significantly more complicated than traditional reconstruction. The core concern is that applying generative AI to human remains raises difficult questions regarding consent and the potential for bias in how a victim is portrayed. If the AI hallucinates features that the victim did not have, it risks misleading the public and potentially hindering the investigation. Privacy and civil liberties groups have echoed these concerns, questioning how such data might be stored, reused, or combined with other datasets in the future.
The Edmonton Police Service has defended the move as a necessary last resort to help identify a victim when all other options had been exhausted. They emphasized that the image underwent internal review and assessment by external experts before being released to the public. However, the incident highlights a growing policy gap. For regulators and policymakers, this case serves as a prompt to establish clearer frameworks. There is currently little guidance on how law enforcement agencies should govern the use of AI-generated imagery in investigations involving sensitive personal data, leaving questions about transparency, retention, and technical standards largely unanswered.
Sources: Edmonton Police Service, Taproot Edmonton
—
By Ji-seo Kim
