Hard Forks and Their Relationship to AI Bug Fixing
Jul 29, 2026 · 18m
Summary
This episode covers Anthropic’s Mythos AI finding Microsoft bugs faster than they can be patched and the fragmentation issues with Google’s Synth ID watermark. It discusses Encore AI’s $30M raise for voice agents trained on sales calls, writers adding typos to evade AI detection, and Sayera’s $1B acquisition of Oasis Security for agent identity management. Brief updates include Google’s Lyria 3.5, Panagram’s text detector, and Martha Stewart’s AI home assistant.
Topics discussed
Intro, AI Box Builder promo, and episode overview
Anthropic Mythos AI outpaces Microsoft bug patching
Google Synth ID watermark durability and fragmentation
Encore AI raises $30M for sales call voice agents
Writers using typos and style breaks to fight AI text
Sierra acquires Oasis Security for AI agent identity
Google DeepMind ships Lyria 3.5 music model
Panagram AI text detection and cat-and-mouse game
Martha Stewart's Hint AI and closing remarks
Listen ad-free on Castria