The CISO's Guide to Google Cloud Next 2026

What to prioritize, what to expect, and what questions to ask before you leave Las Vegas.

Mandalay Bay · April 22–24

Sessions worth your time

Foresite on Stage

Must-Attend Google Sessions

BRK1-087 · Breakout · Anton Chuvakin, Google Cloud + Allie Mellen, Forrester
AI SOC vs. AI in your SOC: What is real today and what is coming ↗
The most important session for CISOs who want to cut through the AI hype. A Google security leader debates a Forrester analyst on what's actually operational versus what's 18 months away. Rare: you'll get real pushback at a vendor conference.

BRK1-085 · Breakout · Spencer Lichtenstein + Wendy Willner, Google Cloud
Agentic defense: Security operations at machine speed ↗
Google's product roadmap for agentic SOC, live demo included. Triage, Investigation, Enrichment, and Response Plan agents working together. Ask about governance controls that prevent false-positive suppression from becoming a liability.

BRK1-103 · Breakout · Nick Bennett + Jurgen Kutscher, Mandiant / Google Cloud
Cybercrime trends: Lessons from the front lines ↗
The annual M-Trends session is consistently the most intelligence-dense content at GCN. This year: how threat actors are leveraging AI, and how organizations are adapting. The Mandiant team has more real IR data than anyone else presenting this week. Don't skip it.

BRK2-205 · Breakout · Google Cloud Offensive Security Services
Securing AI: Real-world lessons and a roadmap for proactive defense ↗
A year of practical lessons securing production AI systems — vulnerabilities found, governance gaps, AI-specific threat modeling in practice. More useful than the conceptual sessions because it's grounded in what actually broke in real deployments.

BRK1-086 · Breakout · Vicente Diaz, Google Cloud + Alexander Pabst, CISO, Allianz SE
Agentic Threat Intelligence: Your new AI security teammate ↗
GTI with a conversational AI interface, demonstrated alongside a real enterprise CISO who's running it in production. The Allianz co-presenter is the credibility signal — this is a peer conversation, not a product demo.

Bonus > Get Hands-On

SECW101 · Capture the Flag Lab · Google Cloud Security
Agentic SOC Experience ↗
Not a competitive CTF — an immersive lab where you work directly with Google's SecOps Alert Triage Agent and Agentic GTI. Send your practitioners here. It's the fastest way to form a real opinion about whether agentic SOC is ready for your environment.

5 questions to ask at the Security Hub

The Security Hub is where the real conversations happen. Don't spend your time watching demos you could watch on YouTube. Come with these.

"Every vendor at this conference will use the word agentic. The question is what it means when the AI gets it wrong."

1. "What does your agentic SOC actually do without a human in the loop?"
   Push on what the AI decides autonomously versus what requires analyst approval. The answer tells you how much they've thought about governance versus how much they've thought about marketing.
2. "How do you handle false positive suppression at scale, and who's accountable when it's wrong?"
   Automation creates liability. If an AI system suppresses an alert that turns out to be real, who owns that decision? Ask for the answer in writing.
3. "What does your MCP implementation look like, and what data leaves my environment?"
   MCP is powerful and relatively new. The data handling implications aren't always obvious. Understand exactly what context your vendor's AI agents are pulling — and where it goes.
4. "Can you show me a real investigation timeline from alert to closure, including every human touchpoint?"
   Not a synthetic demo. A real one. The gap between a demo environment and production is where most agentic security promises fall apart.
5. "What does your SOC 2 report cover, and when was your last audit?"
   If a vendor is asking you to run AI agents in your security environment, they should be able to answer this without hesitation. If they can't, that's your answer.

The AI security risk you can't ignore at a conference about AI

There's an irony in attending a conference about enterprise AI adoption without having audited your own.

The average mid-market organization has dozens of unsanctioned AI tools running across its workforce — LLM integrations, AI coding assistants, shadow agents built by individual teams, third-party SaaS with AI features enabled by default. Most security teams have no visibility into them.

Data exfiltration

Employees paste sensitive data into AI tools that store, train on, or transmit that data outside your control. Most acceptable use policies don't cover tools that didn't exist when they were written.

Shadow agent execution

AI agents built without security review can be granted permissions, make API calls, and take actions in your environment. If you haven't mapped what's running, you can't govern it.

Supply chain exposure

Third-party AI integrations introduce model dependencies you haven't vetted. A compromised or misconfigured model in your stack is a new class of supply chain risk.

The first step is discovery. Before you can govern AI use in your organization, you need to know what's actually running. That's where most security teams are right now — and it's a conversation worth having at the Security Hub this week.

Beyond the security track

While you're at the Security Hub, your engineering and architecture teams should be in the AI platform sessions. The infrastructure being announced there — Gemini Enterprise, Vertex AI Agent Builder, MCP at scale — is what every security announcement is built on. Understanding it gives you better questions to ask vendors.

Four sessions worth routing your team to:

AI Platform Track — Send Your Team