Skip to content

AI Agents

One Boundary for the Agentic Era

with Mark Lechner

Friday 25 September 3:30 PM – 4:00 PM Mainstage

About This Session

This year, an agent carried out an end-to-end attack on real infrastructure, chaining familiar classes of weakness at machine speed. That's the world developers are orchestrating agents in, and orchestration comes with accountability. You've absorbed shift left before; this wave is for how your agents behave. As AI risk becomes enmeshed with supply chain risk, Docker CISO Mark Lechner thinks the answer is a single boundary where agents do their work: isolated execution, trusted content, and controls a security team can verify. Mark’s team is responsible for Docker Hub, which serves billions of pulls a month. In this session he looks at this year's incidents to show what this approach means in practice, and what it changes for defenders. This talk is for anyone who writes code, works with agents, or secures the pipeline they run in. No security background needed.

Topics

  • Agents
  • Agentic AI
  • Cloud Security
  • Containers
  • Governance
  • Security