Let’s talk about securing agentic access at Black Hat
Unless you live on another planet, you’re probably coming to Black Hat with some version of this question: What do we need to do to secure the sensitive systems that AI agents can now access?
There is no shortage of vendors talking about how to discover agents, block the data layer, monitor browser, network or application access. All of that matters.
But agents create a deeper access problem.
An agent is not just a new app or a new workload. It has an identity problem. Sometimes it acts with its own permissions. Sometimes it inherits the permissions of the human, machine or other agent that invoked it. Sometimes it has broader access than the requester. And sometimes nobody can say, with confidence, which identity actually gave it the authority to act.
That makes agentic access a privileged access problem.
If an agent can reach production systems, customer data, cloud environments, ticketing systems, code repositories or internal tools, you need to know more than “is this agent approved?”
As an agent action chain is triggered, you need to know:
- Who or what invoked it
- What identity it is acting as
- What combined permissions it has
- What tools it is trying to use
- What systems it is trying to reach
- What action it is trying to take
- Whether that action should be allowed right now
And at every step of that action chain, you need to have the ability to approve, block or put a human in the loop …based on policy checks taking place at runtime.
Let’s talk about where you are now
Before you can secure agentic access, you need to understand the access model agents are stepping into.
At Black Hat, we’d love to talk through questions like:
- Do you have issues with standing privilege, over-permissioned access or long-lived credentials across users, service accounts, workloads or cloud roles?
- As you think about agents, are you treating them as standalone identities, or are you looking at the full picture – who or what invoked them, what permissions they inherit and what systems they can reach?
- Can you enforce policy across the full action chain before an agent, user or workload takes action in a sensitive system?
Meet with us at Black Hat
Choose what works best:
- 1:1 meeting – Breakfast, lunch, dinner or coffee near the event
- Technical roadmap session – A working session with our team on agentic access, Zero Standing Privilege and runtime policy enforcement
Not going to Black Hat this year? We’ll find time around the event. Just click here.
