Making AI Actually Work in the Enterprise and More RSAC Conference 2026 Interviews - Aamir Lakhani, Camellia Chan, Ely Abramovitch, Jody Brazil, Jim Spignardo - ESW #455

Why It Matters

Enterprises that embed security into AI workflows can avoid costly failures and stay ahead of sophisticated AI‑powered attacks, making robust governance a competitive imperative.

Key Takeaways

• AI workflows fail without integrated security and clear governance
• Fortinet reports surge in AI‑enabled cybercrime in 2026
• X‑PHY’s hardware‑enforced SSD limits AI agent permissions beyond OS
• Legion Investigator uses goal‑oriented AI to replace rigid playbooks
• Zero Trust needs a policy control plane to prevent drift

Pulse Analysis

Enterprises rushing to adopt generative AI often overlook a critical factor: security integration. Jim Spignardo of ProArch argues that AI projects falter when governance, data provenance, and threat modeling are treated as afterthoughts. By embedding security controls directly into AI pipelines—such as model validation, access restrictions, and continuous monitoring—organizations can reduce the risk of model poisoning and data leakage while preserving the agility that AI promises.

The threat landscape is evolving in tandem with AI capabilities. Fortinet’s 2026 Global Threat Landscape Report documents a surge in AI‑enabled ransomware, phishing, and credential‑stuffing attacks that automate reconnaissance and exploit discovery. In response, X‑PHY introduced a hardware‑enforced security layer that isolates AI agents at the SSD level, enforcing immutable permission limits beyond the operating system. This approach curtails the attack surface for autonomous agents, giving enterprises confidence to deploy AI tools without exposing critical assets.

Beyond protection, AI is reshaping how security operations function. Legion Security’s Legion Investigator leverages goal‑oriented AI agents to navigate complex investigations that traditional playbooks cannot handle, delivering analyst‑level reasoning at scale. Meanwhile, FireMon highlights a gap in Zero Trust implementations: the absence of a unified policy control plane that continuously validates enforcement across firewalls, cloud networks, and micro‑segmentation tools. Addressing policy drift with automated validation ensures consistent security posture as infrastructure evolves, cementing AI’s role as both a threat vector and a defensive asset.