AI Agent Security (Jim Miller)

Topic:
AI Agent Security

Description:
As agents become more integrated into our workflows the risk of cyber threats to these agents increases. Recently there have been examples with users running OpenClaw leaking their sensitive data, and being susceptible to indirect prompt injection. We'll explore agent architecture, risks, and how to implement security tooling to mitigate AI threats like: prompt injection, jailbreaks, sensitive data leakage, and misaligned(abliterated) model responses.

Bio:
Jim Miller - Cloud Security Architect, Google - I have spent my career in Cyber Security, and am now focusing on building the next generation of cyber controls for AI workloads and agents

Venue and Food:
We meet at 6:30 pm on the 3rd Thursday of each month. Typically, we meet in person at Paragus IT in Hadley, MA, unless the title notes it as an online-only or hybrid event.

When meeting in person, we appreciate your RSVP so that we can have an accurate pizza count.