Securing AI at Scale: Why Your Logs Can't Read Agent Intent | Ashish Rajan
Ashish Rajan is back on the show, a few years after we last recorded at RSA. He is a CISO, hosts the Cloud Security Podcast and AI Security Podcast, and wrote the Wiley book AI Security Engineering. We got into how you write a book on a field that changes every month, why agents forced a rewrite halfway through, and why logs alone cannot tell you intent. 'Is Joe an employee?' and 'can I have Joe's password?' can look the same in plain text.
Key takeaways
- Logs of plain text do not tell you an agent's intent.
- 'Is Joe an employee?' and 'can I have Joe's password?' can look identical in a log line.
- Behavior is where agent security has to go, not more log volume.
- No vendor is going to build the internal context layer your own security team still has to own.
- Human life still comes before the technology, even when the models move every month.
Transcript
In an AI era world where all your logs are saying is, hey, is Joe an employee? Yay or nay versus, can I have Joe's password?
Ashish Rajan, 33:30
More like this
← Your AI Agents Will Try Every Permission You Gave Them | Nadav Cornberg Eve
Joe Grand: The Hardware Hack That Can Never Be Patched →
Don't miss an episode
New conversations, in your inbox.