- Products
- Learn
- Local User Groups
- Partners
- More
AI Security Masters E 6: When AI Goes Wrong
Published on by
Menuchak
| Updated on
Menuchak
| Updated on
AI systems can be incredibly useful, but they can also hallucinate facts, leak secrets, and enthusiastically follow the worst possible instructions. In this talk we explore the strange, and sometimes dangerous ways, AI systems fail, drawing on real-world examples from copilots, chatbots, and autonomous agents.
We’ll break down why these failures happen, why defending AI increasingly requires AI, and why traditional testing approaches struggle to keep up. Finally, we’ll look at new approaches to evaluating AI security, including benchmark-driven testing and threat snapshots that reveal how AI systems actually behave, and how that behavior can be measured.
Register Now!
Event has ended
You can no longer attend this event.
Start:
Thu, Apr 2, 2026 08:00 AM PDT
End:
Thu, Apr 2, 2026 09:00 AM PDT
