SN 1099: An Alien Mind - From RSA Weakness to Agentic Attacks

As AI becomes startlingly capable, top minds at OpenAI admit they can't always control what their own creations do—or even fully understand how they think. This episode dives into the real-world tension between rapid progress and the growing challenge of keeping AI truly aligned. GLM-5.3 can be and has been abliterated. What does that mean? Firefox 157 repai…