Google DeepMind: The Science of Watermarking AI with Pushmeet Kohli and Jeremy Ratcliffe
In this episode, Professor Hannah Fry speaks with Google DeepMind VP of Science Pushmeet Kohli and Research Scientist Jeremy Ratcliffe about SynthID. The conversation explores how invisible, mathematical watermarks help trace AI provenance and protect against misuse across media such as text, images, and video. The guests explain how the technology is expanding into SynthID Bio to safeguard protein sequence design without impacting biological function. They also examine the broader future of resilience in AI-generated content and biological engineering.
Original sources
Podcast Google DeepMind: The Podcast
Episode:From deepfakes to DNA: the science of watermarking AI ↗
YouTube Google DeepMind
Video:From deepfakes to DNA: the science of watermarking AI ↗
AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.