Gill Verdon Claims Shadow Integration Blocks Adversarial Hacks

Gill Verdon, founder and CEO of Extropic, shared a post explaining shadow integration. He described it as making latent variables visible nodes so AI models avoid adversarial input hacks. The statement positions the change as a way to expose hidden variables and block attacks. Verdon leads a startup focused on energy-efficient AI hardware using thermodynamic computing. The tweet also noted he is known as Guillaume Verdon-Akzam. The post offers no code, examples, or results. It stands as his individual claim about the technique without further confirmation from other sources in the packet.