Your company grew this year, just not the way your org chart says. Forcepoint CEO Ryan Windham opens AWARE with the workforce nobody interviewed, nobody onboarded and nobody can slow down. His argument is uncomfortable: no agent has ever exceeded its authority. It exceeded your intent, and there is no permission for that. Ryan makes the case that security has to stop governing authority and start governing the use, and shows what it takes to command AI to put your organization’s best intelligence to work. Plus a first look at what Forcepoint ships next.

Here's the uncomfortable truth about agentic AI: there's no best practice yet. Every enterprise is running the same experiment on the same frontier, inspired by what these systems can do and accountable for everything that goes wrong. Futurist Ian Beacraft, one of the most sought-after voices on AI and the future of work, and Forcepoint's David Giambruno cut through the fog together. They unpack what actually changes when agents stop being tools and start being teammates — a modern workforce that moves data at machine speed and behaves in ways that would trip every alert you have if a human did the same. From sanctioned copilots to shadow tools to autonomous agents, they get to the insight most teams are missing: the risk isn't confined to any one tool, it follows the data across all of them. Expect a candid, tech-forward conversation and a way to read the signals of change before the risk arrives.

What you'll gain

• A clear picture of why the pre-agentic security model breaks when non-human actors do the work at machine speed.
• A prioritization checklist that makes clear why the agent or model is the commodity and your data is the real threat surface, including the organizational data most enterprises have never secured.
• A mindset for leading where there's no playbook: treating AI as adaptive R&D, so your team moves with curiosity and realistic goals instead of panic.

AI governance and data security as one continuous, adaptive motion, from the data layer on up. Bakshi Kohli, Salah Nassar and Saritha Chadalavada walk through the latest advances on discovery of unsanctioned AI and sanctioned AI across apps, browser extensions and endpoint agents. See real-time prompt and response inspection that protects sensitive data as it moves through AI interactions. Meet ARIA, the Adaptive Risk Intelligence Assistant that turns natural-language intent into machine-ready governance and protection policy, across channels. And see single-console AI oversight across ChatGPT Enterprise, Claude for Enterprise and Microsoft Copilot, plus AI-powered classification across structured and unstructured data, all from one platform with one policy framework for every AI interaction.

Enterprises spent decades moving data to the cloud. They are spending days connecting AI to it, and the pace of that second shift is nothing like the first. Forcepoint President of Go-to-Market Rick Hanson sits down with AWS Principal Applied AI Architect Neelam Koshiya to work through what agentic AI does to the shared responsibility model, drawing on what AWS sees across its customer base. Along the way: shadow AI, agent-to-application data flows, where Amazon Bedrock fits and the question every enterprise eventually faces — who decides when an agent goes to production.

What you'll gain

• Shared responsibility extends into agentic security. Hear how the familiar model changes when agents act on their own: what the cloud provider secures, what a security vendor enforces, what stays with your security, IT and compliance teams — and what each part gives you in return.
• Non-human identities change the math. Learn what agents are actually doing inside customer environments, why volume alone breaks identity-centric controls and where the durable control point moves when new actors appear faster than any review cycle can approve them.
• Ownership divides. Accountability doesn't. Understand how guardrails, data controls and services like Bedrock help you hold the line — and why you still own compliance.

AI security and data security aren't two problems. They're a package deal. Most organizations are still funded, staffed and governed as though they aren't. Assurance is where it shows first. In July 2026 Oracle shipped 1,449 patches in a single quarterly update, nearly triple its record, most of them found by AI. As a result of patch overload, it then moved to monthly updates, because quarterly couldn't keep up. The certifications your customers rely on were built for a slower cycle. Roland Cloutier and Ronan Murphy discuss running an AI security program when the frameworks and guardrails it was built on keep changing. How did you red team your own code? What has AI touched, and can you prove it stayed protected? Your customers and constituents are already asking. A trust program is how you answer with evidence, not assurances.

What you'll gain

• Why AI security and data security can no longer be funded, staffed or governed as separate programs.
• What to stand up now: AI scorecards, assurance programs and agentic review.
• How to answer "how did you red team your own code?" before a customer asks it.
• Achievable steps a security or compliance leader can start on right away.