Talks
Practitioner-grounded talks on AI security
Anchored in systems I run, not survey papers. Slides, demos and write-ups go up here after each one.
Contributor, OWASP Top 10 for Agentic Applications 2026 · CISSP · PhD · 15+ years enterprise security
Where I'll be next
7–8 September 2026
Park Plaza Victoria, London
fwd:cloudsec Europe 2026
Blast Radius for AI Agents: Containment When Prompt Filters Fail
Talk page, slides and takeaways →17–18 September 2026
RAI Amsterdam
AGNTCon + MCPCon Europe · MCPCon
What Breaks When Agents Get Tools: MCP Security in Production
Talk page, slides and takeaways →Attending either? Find me afterwards. The talk pages above are where the slides, the demo and the full write-up land.
For organisers
Talk topics
Each topic is a 30 to 45 minute talk that can be delivered as a conference session or expanded into a half-day workshop.
Securing Agentic AI Systems: A Field Engineer's Guide
Walkthrough of the OWASP Top 10 for Agentic Applications 2026 with live demonstrations of prompt injection, tool poisoning, and confused deputy attacks against representative agent architectures.
Best for: technical conferences, OWASP chapters, security engineering teams.
The MCP Security Disaster: What Goes Wrong When Agents Get Tools
Specific deep dive on Model Context Protocol attack patterns. Tool poisoning, meta-context injection, cross-server attacks. Practical mitigations.
Best for: AI engineering audiences, infrastructure security teams.
AI Act Article 15 in Practice: Cybersecurity Requirements for High-Risk AI
The technical control mapping behind the EU AI Act's cybersecurity requirements. What the article actually requires of engineering teams.
Best for: compliance and governance audiences, IAPP chapters, regulated-industry security teams.
Threat Modeling for Agent Architectures
Hands-on workshop format. Extends STRIDE for agentic systems. Participants threat-model a representative system live.
Best for: engineering teams, internal training.
Previously
OWASP Stockholm: MCP Security, One Year In
OWASP Stockholm Chapter · May 2026
A debrief from a year of hands-on Model Context Protocol security work: where servers trust callers by default, the attack patterns that show up most often, and the mitigations that actually hold.
Read the debrief →What audiences get
- Slides and supporting material delivered before the event so AV and accessibility checks can happen properly
- A demo environment available for hands-on segments (no production data, runs in a sandbox)
- Time set aside for Q&A. The most valuable part of a talk is usually the questions.
- Follow-up handouts for participants who want to go deeper
Corporate training
For private engagements (your own engineering org, security team, or partner audience), training is delivered as a 90-minute executive briefing, a half-day workshop, or a private on-site. Get in touch and we'll work out the right format for your team.
Booking
| Type | Format | Booking lead time |
|---|---|---|
| Conference talk | 30 to 60 minutes, on stage | 6 months ahead is best |
| Private corporate training | Half or full day, on-site or remote | 8 to 12 weeks |
| Online workshop or webinar | 60 to 120 minutes | 4 to 6 weeks |
Based in Gothenburg, Sweden. Comfortable traveling within Europe. One to two long-haul trips per year for major events. Honoraria expectations standard for senior practitioner speakers; exact terms depend on event size and format.
Submit a speaking request →