Who Has the Kill Switch? Runtime Security for Autonomous AI Agents

Grab some coffee ☕, pull up a chair, and join us for an evening of agentic security discussions!

Giving AI agents tool access, database roles, and autonomous execution power is the fastest way to 10x your output—and the fastest way to accumulate silent, catastrophic 🧨 operational risk.

While traditional security tools monitor human logins or scan static code before deployment, what happens when an autonomous agent executes 20 lateral tool calls across your infrastructure in under a minute?

On October 15th, AI Club Nordics welcomes Anton Renmark (Co-Founder of Hejdar) to tear open the hood on Agent Security & Runtime Control.

We’ll look past basic prompt guardrails and dive into the architecture needed to decide what agents are allowed to do before they execute:

🛡️ Inline Policy Interception: Evaluating every tool call in sub-second time.
🛑 Runtime Kill Switches: Fail-closed design patterns to instantly freeze out-of-bounds agent loops.
📝 Deterministic Governance: Immutable audit logging built for enterprise security and EU AI Act compliance.

Come for the raw code breakdowns, stay for the social networking, and vibe with your fellow club members!

🗓️ Thursday\, October 15 | 17:30 – 20:00
Embassy House Coworking Medborgarplatsen
📍 Östgötagatan 12, 116 25 Stockholm