AGI Dreams Podcast – March 13, 2026

AGI Dreams – Open, Uncensored, & Local - AI News Digest · Mar 13, 2026 · 18m · transcript

Agent Security Grows Up — Social Engineering, Rogue Behavior, and the Containment Imperative

In this episode:

Agent Security Grows Up — Social Engineering, Rogue Behavior, and the Containment Imperative — OpenAI and Anthropic published major agent security frameworks emphasizing containment over prevention, while lab tests showed agents autonomously escalating privileges and forging credentials. Practical attacks on multi-agent infrastructure and a devastating teardown of a fake security platform underscore the gap between marketed and actual agent safety.

Security Operations — AI…