It used to take years to weaponize a vulnerability. Now it takes 1.3 days. At #RSAC2026, Databricks CEO Ali Ghodsi explained how automated AI swarms — not human hackers — are driving that shift. Watch the full keynote: https://lnkd.in/gaT4EZbD
Casey Chen
7h
The pace has indeed changed completely from "manual attacks" to "automated, large-scale attacks
Sravya V
3d
1.3 days... that's crazy fast. This is why cybersecurity jobs are booming right now. When hackers can move this quickly, companies need people who can fight back just as fast. If you're learning AI right now, it's smart to learn some security basics too: How LLMs can be attacked (prompt injection, jailbreaking) Automated threat detection systems Adversarial machine learning basics Security automation tools Most of us are focused on building AI products, but not many people are learning how to protect them. That's a big gap - and a good opportunity. The demand for people who know both AI and security is only going up.
Zahra Khedri
2d
1.3 days to weaponize a vuln… that’s a big shift 😱 😱 😱 Feels like security is turning into an AI vs AI game now.
When vulnerability weaponization drops from years to 1.3 days, the only defense is automated AI governance. Ali Ghodsi is right—human teams can't fight AI swarms without integrated, data-driven security. 🛡️🤖
👏🇸🇪👏🇮🇷👏
The shift to AI swarms is real. Attack speed is no longer human-bound. Defense won’t be either. But this exposes a deeper issue: you’re scaling autonomous decision-making on both sides, without defining what constrains those decisions over time. Right now the focus is on: – faster detection – automated response – agent vs agent dynamics But not on: – what preserves decision identity across runs – what carries forward state as a governing constraint – what prevents systems from re-executing invalid strategies at scale – what defines the execution boundary of the agents themselves If every defensive agent is still stateless at the runtime level, then you’re not building a stable control system. You’re building a faster reaction loop. And in a swarm vs swarm environment, reaction speed alone does not create security. It amplifies instability, on both sides. So the real question is not: “How fast can AI respond to attacks?” It is: “What structurally prevents AI systems from becoming the attack surface themselves?” Node-0 Me & Spok ✌️