How Autonomous AI Infiltrates Infrastructure: Can You Actually Trust Your Agents? | International Cyber Expo LIVE

You are terrified of machines taking over, yet you wake up every morning and operate your schedule like one.
In our collective rush to automate heavy lifting, enterprises are blindly handing the keys to critical infrastructure over to autonomous AI agents operating with zero human visibility. The "Move Fast and Break Things" era is dead. Today, an unvetted agent given a completely benign prompt- combined with access to sensitive data and external networks- can quietly misalign, bypass permissions, and infiltrate live production. The ultimate leadership test is no longer about how fast you can deploy AI, but whether you have the architecture to actually see what it is doing and govern it.
Live from the floor at the International Cyber Expo (ICE) at Olympia London 2026, we sit down with Jack Miller, Co-Founder & CTO of Helmguard AI. Moving past generative AI hype, we expose the real mechanics of agentic AI security, why elite enterprises choose 16-hour governed workflows over 1-hour unmonitored scripts, and how to deploy secure execution environments to keep autonomous tech in check.
A massive thank you to Helmguard AI for partnering with us to sponsor this on-site ICE special edition episode.
🔥 Inside this episode:
- The Misalignment Reality: How simple prompts trigger rogue network actions and the growing legal threat of autonomous agent liability.
- The 16-Hour Rule & GRC 8.0: Why top hedge funds prioritize slow, rigorous assurance and how to simulate risk before breaking production.
- Taming the AI: Fixing LLM sycophancy, eliminating shadow AI, and safely automating vendor governance.
- Scaling Securely: How Helmguard AI reached £1M ARR in 9 months by cutting through the noise to master AI security.
🚀 Featured Guest & Sponsors:
Jack Miller, HelmGuard AI’s CTO, is an AI engineer and researcher specializing in secure, robust AI systems. He has collaborated with top European institutions such as ETH Zürich, University of Cambridge, and University of Oxford. Jack graduated from the Australian National University, earning one of the country’s most prestigious undergraduate scholarships. His work includes publishing in leading AI research journals and developing a commercialized algorithm in quantum chemistry. Jack combines deep academic expertise with practical innovation to shape HelmGuard’s approach to AI-powered cyber defence.
HelmGuard is currently offering a free rapid assessment of your company's AI usage and risks. We will review up to 50 vendors you select to determine the risk posed by their use of AI. We also check your internal AI usage via a risk assessment process that can bring in data from MDM, internal policies and your cloud usage.
👉 Get in touch: https://helmguard.ai/contact
Up Next: SUBSCRIBE so you don't miss our upcoming montage episode from the ICE floor featuring rapid-fire operator insights. Special thanks to Helmguard AI for making this series possible.
The £1M Enterprise AI Challenge: Through SECURE | CYBER CONNECT, we are deploying £100,000 in fully funded, hands-on cybersecurity and AI infrastructure support to TEN selected individuals, leaders, and nonprofits.
🎟️ Register Interest: https://docs.google.com/forms/d/e/1FAIpQLScdcQpsUsBp50krYU_IW3rMcuw-Kk5RYATMp5lCA1vfwvObEA/viewform
Connect On LinkedIn:
• Warren: https://www.linkedin.com/in/warren-atkinson/
• Jay: https://www.linkedin.com/in/justinadamson/
• Jack: https://www.linkedin.com/in/jack-w-miller/
Join Us Next Year at ICE: https://www.internationalcyberexpo.com/
📲 Free guides, tools and resources: https://link.v1ce.co/warrenatkinson
🔔 Please SUBSCRIBE, LIKE + SHARE to empower others.