Arachnid: Hunting the web’s darkest secrets Ep 2: The Solution
May 27, 2025
Guest
Leanna McDonald
Guest
Microsoft representative (Tim Cranston / legal/tech contact)
Guest
Hany (Hanny) - Professor / PhotoDNA inventor
Leanna McDonald, director at the Canadian Center for Child Protection who leads Project Arachnid. Tim Cranston, former Microsoft legal/tech contact who coordinated early PhotoDNA work. Hany, the academic inventor of PhotoDNA. They discuss how PhotoDNA was built and deployed quickly, why known-image matching matters, the scale of the global problem, and why the job remains unfinished.
AI Snips
Chapters
Transcript
Episode notes
Discovering Her Abuse Through Photos On A Kitchen Table
- Sage learned from a detective that images of her childhood abuse existed online and saw cleaned-up photos laid out on her kitchen table.
- Seeing the pictures triggered flooding memories, near fainting, and the realization that others had viewed and shared those videos.
Solve CSAM By Matching Known Images Not Classifying Everything
- Hany reframed the problem: instead of detecting any abusive image at scale, match known illegal images already identified as CSAM using hashes.
- That approach led to PhotoDNA within a year and enabled platforms to find vast sets of related material cheaply.
PhotoDNA Built By Two Engineers And A Dozen Lawyers
- PhotoDNA was built by a tiny team: Hany, one Microsoft engineer, and many lawyers, producing a working prototype in nine months and a deployable tool in a year.
- The tool worked so well that Microsoft released it in the cloud and it became widely adopted across large platforms.