Self-Replicating Prompt Injections Are Here

Prompt injection gets much more serious when AI agents can carry the attack forward themselves. Self-replicating instructions point toward a future where securing agentic systems may require solving several safety problems at once. Check out the full episode now #short