Presentation
Security Considerations for Implementing GenAI on z/OS
Contributor
Event Type
Technical Session
Machine Learning/AI
Security and Compliance
z/OS Systems Programming
Soft Skills
All Audiences
Modernize Data Management and Analytics
TimeTuesday, August 1810:30am - 11:30am EDT
LocationRoom 308
DescriptionBefore you roll GenAI on the mainframe out, are you asking the right security questions?
This session cuts through the excitement and gets practical. As organisations begin integrating Generative AI capabilities into their z/OS environments, a new set of security challenges comes with it and the stakes on the mainframe are too high to learn lessons the hard way.
We'll take a hands-on, no-nonsense look at what you actually need to think about before and during a GenAI implementation. From ensuring your existing RACF controls extend effectively to AI workloads and data access, to understanding the very real risks of prompt injection and model manipulation This session will give system programmers and security teams a shared language and a practical framework for getting it right.
You don't need to be an AI expert to follow along. If you're responsible for keeping z/OS secure, this session is built for you.
What we'll cover:
- Extending RACF and access control principles to GenAI workloads
- Understanding prompt injection - what it is, why it matters on the mainframe, and how to mitigate it
- Practical steps to harden a GenAI implementation without killing its usefulness
- Where the biggest risks are hiding and how to get ahead of them
This session cuts through the excitement and gets practical. As organisations begin integrating Generative AI capabilities into their z/OS environments, a new set of security challenges comes with it and the stakes on the mainframe are too high to learn lessons the hard way.
We'll take a hands-on, no-nonsense look at what you actually need to think about before and during a GenAI implementation. From ensuring your existing RACF controls extend effectively to AI workloads and data access, to understanding the very real risks of prompt injection and model manipulation This session will give system programmers and security teams a shared language and a practical framework for getting it right.
You don't need to be an AI expert to follow along. If you're responsible for keeping z/OS secure, this session is built for you.
What we'll cover:
- Extending RACF and access control principles to GenAI workloads
- Understanding prompt injection - what it is, why it matters on the mainframe, and how to mitigate it
- Practical steps to harden a GenAI implementation without killing its usefulness
- Where the biggest risks are hiding and how to get ahead of them