The Internet Has No Off Switch: AI’s Security Debt, ICANN Lessons & the Scam Tsunami

The internet was built for a handful of universities, with no security where it seemed unnecessary. We’re now making the same mistake with AI, only faster. The Problem:  Once a technology spreads, it’s “humpty dumpty.” You can’t put security back in. Brian Cute has watched this happen from the inside: at a submarine cable carrier during the dot-com crash, at Network Solutions when it was the only domain seller on Earth, and inside ICANN. This Cybersecurity Awareness Month, he explains why the rush to adopt AI is repeating the internet’s original design mistake, and what history says we should do instead. Who is this for:  Security architects, GRC and compliance professionals, policy watchers, and technology leaders adopting GenAI and agentic AI without a clear risk model. Summary:  The Original Sin Vint Cerf and the early internet’s designers built for scale within a closed group of universities, and some security features were left out. When the network went global, criminals found those gaps. “Getting everybody to adopt a security fix is really, really hard, next to impossible,” Brian says. Everyone rushing to deploy AI, he warns, is watching “a similar movie, yet again.” Take the Word “Governance” Away ICANN worked because the DNS is a scarce, single-operator namespace. AI has none of those attributes. Brian argues the real work is building collaboration mechanisms: a defined problem, a shared outcome, and clear roles for government, industry, civil society, and engineers. Justin adds that existing privacy, fraud, and IP law already covers much of what’s needed. Bring a Tool, Then Get Out of the Way Brian’s lesson from rural communities: “All of the knowledge that is required to address these problems already exists.” Outsiders should bring a tool that’s fit for purpose and let the community adapt it. 📩 Subscribe to Secure Talk: 🔐 Brought to you by Strike Graph: www.strikegraph.com Chapters 00:00 Introduction and guest background 04:22 Starting the conversation on internet evolution 09:27 Lessons from the domain name system and security challenges 12:27 AI's impact on cybercrime and automation 17:30 Scaling cybersecurity tools for small organizations 21:47 Measuring security impact and behavior change 25:35 AI as a tool for protection and learning 36:08 Governance of AI and internet infrastructure 40:07 Collaborative approaches to AI governance 43:58 Closing remarks and future outlook 📚 SOURCES Global Cyber Alliance: https://globalcyberalliance.org IETF: https://www.ietf.org Microsoft Digital Defense Report 2025: https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Microsoft-Digital-Defense-Report-2025.pdf CISA Cybersecurity Awareness Month: https://www.cisa.gov/cybersecurity-awareness-month ICANN: https://www.icann.org Model Context Protocol: https://modelcontextprotocol.io ABOUT THE GUEST Brian Cute is CEO of the Global Cyber Alliance and former CEO of Public Interest Registry, operator of .org. He spent decades in ICANN policy work, beginning as director of policy at Network Solutions. Hosted by Justin Beals, founder and CEO of StrikeGraph. #CybersecurityAwarenessMonth #InternetGovernance #AIGovernance #AISecurity #SecureTalk