In 2026, securing AI-generated code feels like a solved problem. The threats that replaced it are not.

This session examines converging risks that are actively reshaping the enterprise security landscape: the erosion of developer judgment through uncritical AI adoption; an agentic attack surface expanding far faster than the tools designed to protect it; encrypted secrets already being harvested for future decryption; and a geopolitical price war quietly normalizing foreign AI models in infrastructure where trust has never been established.

Drawing on recent exploit research and real-world adoption data, this talk delivers a practical, defense-in-depth framework for organizations that are serious about keeping their pipelines secure as AI becomes load-bearing infrastructure.

Attendee Takeaways:

- Recognize how AI adoption patterns are creating exploitable vulnerabilities at the human layer, not just the technical one
- Understand the expanding attack surface of agentic systems and what defenders need to prioritize right now
- Make the case internally for post-quantum cryptography investment before the window closes
- Evaluate the hidden security costs of commodity AI models before they become a supply chain liability