Its scary to see that these vulnerabilities aren’t just a one off issue but more a symptom of how authorization works today. Seems similar to the confused deputy problem which im surprised still hasnt been solved for…
Interesting to experiment with. As AI agents get faster and cheaper, having good governance around their actions becomes more important to cut down on runaway agentic actions and consequences
Its scary to see that these vulnerabilities aren’t just a one off issue but more a symptom of how authorization works today. Seems similar to the confused deputy problem which im surprised still hasnt been solved for…
Interesting to experiment with. As AI agents get faster and cheaper, having good governance around their actions becomes more important to cut down on runaway agentic actions and consequences