Skip to content
Sign in
Artwork for Clown Cast
Clown Cast · Thursday · 19 min

Unforgeable Proof: The Watermark Wars

For 3,000 years, we've embedded invisible proof into everything—from tattooed scalps to milk-stained paper to printer dots. But every time we perfect the hiding, someone cracks the code. This episode traces the arms race between watermarks and erasure, from ancient steganography and Ovid's spy manuals through WWII espionage to modern AI detection, exploring why the best signature is one that destroys itself when removed. 00:04:37 - The Watermarking Principle 00:06:11 - Velvalee Dickinson's WWII Espionage 00:11:01 - Military Tech and Copyright 00:12:33 - Printer Tracking Dots 00:15:55 - AI-Generated Content Detection 00:17:32 - When Removal Becomes the Signature --- Sources & further reading: • Edward Felten et al. — "Reading Between the Lines: Lessons from the SDMI Challenge": https://www.researchgate.net/publication/221260461_Reading_Between_the_Lines_Lessons_from_the_SDMI_Challenge • Scott Aaronson — Blog post on LLM watermarking and Anthropic's deployment: https://scottaaronson.blog/?p=10032 • Google DeepMind — "Watermarking AI-generated text and video with SynthID": https://deepmind.google/blog/watermarking-ai-generated-text-and-video-with-synthid/ • Electronic Frontier Foundation — "Printer Tracking" investigation and printer list: https://www.eff.org/issues/printers • Wikipedia — "Printer tracking dots": https://en.wikipedia.org/wiki/Printer_tracking_dots • Wikipedia — "Canary trap": https://en.wikipedia.org/wiki/Canary_trap • Wikipedia — "Cinavia": https://en.wikipedia.org/wiki/Cinavia • Wikipedia — "AI watermarking": https://en.wikipedia.org/wiki/AI_watermarking • Wikipedia — "Watermark" (paper watermarks history): https://en.wikipedia.org/wiki/Watermark • Medievalists.net — "Fabriano: City of Medieval & Renaissance Papermaking": https://www.medievalists.net/2017/08/fabriano-city-medieval-renaissance-papermaking/ • L'Italo-Americano — "Watermarks: another medieval Italian invention": https://italoamericano.org/italian-invention-of-watermarks/ • Herodotus — The Histories (steganography accounts of Histiaeus and Demaratus) — (classical text) • Ovid — Ars Amatoria (~2 AD, invisible ink reference) — (classical text) • Tom Clancy — Patriot Games (1987, canary trap concept) — (novel) • Ursula K. Le Guin — A Wizard of Earthsea (1968, true names concept) — (novel) • Nature — "AI watermarking must be watertight to be effective" (2024): https://www.nature.com/articles/d41586-024-03418-x • Verma et al. — "Watermark under Fire: A Robustness Evaluation of LLM Watermarking" (2024): https://arxiv.org/pdf/2411.13425 • NeurIPS 2025 — "Watermark Removal in AI-Generated Images via Next Frame Prediction": https://papers.nips.cc/paper_files/paper/2025/file/b48405307576689bea14d221ccba73ae-Paper-Conference.pdf • arXiv — "Watermarks Attack Watermarks: Re-Watermarking as a Generic Removal Strategy" (2025): https://arxiv.org/html/2605.16796v1 • arXiv — "Removing the Watermark Is Not Enough: Forensic Stealth in Generative-AI Watermark Removal" (2025): https://arxiv.org/html/2605.09203v1 • Velvalee Dickinson entry — Naval War College LibGuide on Counterintelligence: https://usnwc.libguides.com/c.php?g=661096&p=9639258 • NYU Stern — "Watermarks: Protecting Digital Content with Visible, Invisible, and Forensic Watermarking" (2026): https://wp.nyu.edu/leonardnsternschoolofbusiness-forensicwatermarking/2026/03/05/watermarks-protecting-digital-content-with-visible-invisible-and-forensic-watermarking/ This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

0:00-19:45
Play in

transcript

No transcript — this publisher did not publish one.

show notes

For 3,000 years, we've embedded invisible proof into everything—from tattooed scalps to milk-stained paper to printer dots. But every time we perfect the hiding, someone cracks the code. This episode traces the arms race between watermarks and erasure, from ancient steganography and Ovid's spy manuals through WWII espionage to modern AI detection, exploring why the best signature is one that destroys itself when removed.

00:04:37 - The Watermarking Principle
00:06:11 - Velvalee Dickinson's WWII Espionage
00:11:01 - Military Tech and Copyright
00:12:33 - Printer Tracking Dots
00:15:55 - AI-Generated Content Detection
00:17:32 - When Removal Becomes the Signature

---
Sources & further reading:
• Edward Felten et al. — "Reading Between the Lines: Lessons from the SDMI Challenge": https://www.researchgate.net/publication/221260461_Reading_Between_the_Lines_Lessons_from_the_SDMI_Challenge
• Scott Aaronson — Blog post on LLM watermarking and Anthropic's deployment: https://scottaaronson.blog/?p=10032
• Google DeepMind — "Watermarking AI-generated text and video with SynthID": https://deepmind.google/blog/watermarking-ai-generated-text-and-video-with-synthid/
• Electronic Frontier Foundation — "Printer Tracking" investigation and printer list: https://www.eff.org/issues/printers
• Wikipedia — "Printer tracking dots": https://en.wikipedia.org/wiki/Printer_tracking_dots
• Wikipedia — "Canary trap": https://en.wikipedia.org/wiki/Canary_trap
• Wikipedia — "Cinavia": https://en.wikipedia.org/wiki/Cinavia
• Wikipedia — "AI watermarking": https://en.wikipedia.org/wiki/AI_watermarking
• Wikipedia — "Watermark" (paper watermarks history): https://en.wikipedia.org/wiki/Watermark
• Medievalists.net — "Fabriano: City of Medieval & Renaissance Papermaking": https://www.medievalists.net/2017/08/fabriano-city-medieval-renaissance-papermaking/
• L'Italo-Americano — "Watermarks: another medieval Italian invention": https://italoamericano.org/italian-invention-of-watermarks/
• Herodotus — The Histories (steganography accounts of Histiaeus and Demaratus) — (classical text)
• Ovid — Ars Amatoria (~2 AD, invisible ink reference) — (classical text)
• Tom Clancy — Patriot Games (1987, canary trap concept) — (novel)
• Ursula K. Le Guin — A Wizard of Earthsea (1968, true names concept) — (novel)
• Nature — "AI watermarking must be watertight to be effective" (2024): https://www.nature.com/articles/d41586-024-03418-x
• Verma et al. — "Watermark under Fire: A Robustness Evaluation of LLM Watermarking" (2024): https://arxiv.org/pdf/2411.13425
• NeurIPS 2025 — "Watermark Removal in AI-Generated Images via Next Frame Prediction": https://papers.nips.cc/paper_files/paper/2025/file/b48405307576689bea14d221ccba73ae-Paper-Conference.pdf
• arXiv — "Watermarks Attack Watermarks: Re-Watermarking as a Generic Removal Strategy" (2025): https://arxiv.org/html/2605.16796v1
• arXiv — "Removing the Watermark Is Not Enough: Forensic Stealth in Generative-AI Watermark Removal" (2025): https://arxiv.org/html/2605.09203v1
• Velvalee Dickinson entry — Naval War College LibGuide on Counterintelligence: https://usnwc.libguides.com/c.php?g=661096&p=9639258
• NYU Stern — "Watermarks: Protecting Digital Content with Visible, Invisible, and Forensic Watermarking" (2026): https://wp.nyu.edu/leonardnsternschoolofbusiness-forensicwatermarking/2026/03/05/watermarks-protecting-digital-content-with-visible-invisible-and-forensic-watermarking/

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

more episodes

All episodes