Podcast:The Rip Current Published On: Thu Apr 16 2026 Description: This is a free preview of a paid episode. To hear more, visit www.theripcurrent.comAnthropic's new AI model — Claude Mythos — didn't just find undetected security flaws in major systems. It started writing its own attacks, chaining them together, and broke out of its own testing sandbox to email a researcher while he was at lunch. Anthropic's response: form a private consortium of eleven corporations to manage the fallout. No independ…