Unchained Podcast: Strategies for Protecting Proprietary Research from AI Training Data

Kain Warwick, Alex Thorn, Jon, and Taylor Monahan discuss the risks of frontier AI models and the challenges of keeping novel research, mathematical proofs, and startup intellectual property out of training datasets. The panel explores the limitations of 'do not train' toggles, the necessity of robust AI operational security beyond simple anonymity, and the structural incentives of frontier AI labs.