AI & ML in Identity Security

[alind141529]

Hi All,

Let's discuss how the AI and ML will disrupt the Identity security and what cons would it bring with it.

[amy_shillinglaw]

AI identities bring some of the same and some unique risks compared to human ones. I don't feel like AI will disrupt Identity Security if we plan and execute responsibly and ethically, putting in the proper guardrails from the beginning, then using SailPoint's solutions to have visibility of our non-human accounts and govern them.

Of course, shadow agents may, at times, get away from us, or some organizations may not invest in controls equally. But if we think it through, plan well, and have safeguards in place to detect the risks, the entire identity landscape can evolve more safely and securely.

[alind141529]

Agree with the challenge of scalability, however AI can be trained over a period of time to address the issues around situational awareness. ML helps it to constantly evovle.

[paul_patti]

@alind141529 so true! Training AI to perform routine functions in Identity Security will indeed be interesting.

In 2019 I attended a seminar at UC Berkeley hosted by NVidia that showed how a dog door app could be trained to let in a dog, but not a cat or any other animal. And now they have gone even further using AI to train dogs, you can read about it here!

There will probably be many identity security functions related to pattern recognition AI will be able to help with, and so much more!

[paul_patti]

@alind141529 great question! For starters, the need to address many times the number of human identities we have today will be huge - so the challenge of scalability!

Another key challenge will be context. AI is great about knowing exactly what to do, but situational awareness is not its strong suit. For example, how can we train AI to produce more accurate and helpful information to support IGA decisions - especially as standards, rules, and regulations evolve?