Databricks exec: Enterprise AI security in 2026 still isn't where it needs to be

TechCrunch Disrupt 2026 is dragging the AI conversation out of the demo room and into the boardroom. The AI Stage, running October 13–15 in San Francisco, isn’t about what’s cool—it’s about what’s broken. And plenty is broken.

Databricks co-founder Arsalan Tavakoli is set to deliver a reality check on enterprise security. The core problem isn’t a lack of AI tools; it’s that autonomous AI agents are now making decisions inside sensitive systems at speeds that make traditional security frameworks look like a horse and buggy. Tavakoli’s session promises to map out the architecture that separates a trustworthy deployment from a catastrophic liability. It’s a conversation about observability and governance that most founders are still avoiding.

On the product side, AI pricing is having its own existential crisis. With models becoming commoditized, the old rules for SaaS margins are collapsing. The stage will force founders to confront the question they dread: how do you price a product when the core technology is trending toward free? Amit Jain from Luma AI and Dean Leitersdorf from Decart will also hit the stage to move the video AI discussion past shiny object demos and into what real-time, physical reasoning actually unlocks for businesses.

Then there’s the job market. Kareem Amin from Clay will break down the rise of the “GTM Engineer,” a role that didn’t exist two years ago and now has independent practitioners building seven-figure businesses. It’s a signal that AI isn’t just automating old work—it’s creating entirely new career tracks that require a fluency most growth teams lack. The event runs alongside Startup Battlefield and a 10,000-person network of VCs and founders, with a $300 discount window closing soon.