Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Security investment decisions are always, at some level, conversations about risk quantification: how much exposure does the organization actually carry, what does it cost to reduce that exposure, and what's the cost of not acting at all. For agentic AI security, these questions are answerable once the right measurement is in place.

As AI agents operate with increasing autonomy in enterprise production, the sheer volume of security telemetry they generate creates an unprecedented scaling challenge. Human oversight remains required in many enterprises, while legacy triage tools simply can't keep up. To help security teams cut through the noise, Zenity’s AI group built the Blue Agent.

Anthropic's Inference Hooks update lets security teams validate Claude's tool calls before they run. Zenity supports it across Claude's surfaces, so teams can approve write-capable connectors, Model Context Protocol (MCP) servers, and wider Claude Code rollouts with policy on each action.

The agentic AI security conversation has, for good reason, spent most of its energy on the agents an organization builds or deploys directly: what they're authorized to do, what data they can reach, how their behavior is monitored. I think that focus is necessary and also increasingly incomplete, because the fastest-growing risk surface in agentic AI right now isn't the agents themselves.

Japan’s Ministry of Finance reports that AI adoption in large local enterprises has surged to 89%. Companies are progressively choosing to incorporate AI agents into their workflows, building with Microsoft Copilot Studio, ChatGPT Enterprise, Claude, and a myriad of other platforms, all living across a variety of environments with autonomy, privilege, and minimal supervision.

Your team just hired someone who never sleeps, reads everything, and takes work from whoever asks. That’s the elevator pitch for Claude Tag, and on the face of it, it’s hard to see why anyone would push back on the new hire. Enterprises will want this in every workspace. On-demand status updates, first drafts, streamlined triage; the work that quietly eats a week now collapses into the team conversation.

Most of the industry discourse on agentic AI risk has settled into a comfortable framing: agents get attacked the way models get attacked, through some form of prompt manipulation, and the fix is a better guardrail. I think that framing is dangerously incomplete, and I want to walk through five specific scenarios that make the case directly rather than abstractly. Two of them are manipulation. One exploits trust between agents rather than any single agent's behavior.

Zenity Labs is publicly launching AI Total: a free service that runs an AI agent skill in a sandbox and tells you what it actually did, before you let it near your agents.

Ask AI to Choose a prompt Write a TLDR of this post Explain the security risk Summarize what CISOs should know Many conversations about AI agent risk over the past year start from the same unspoken assumption: something bad happened because someone or something manipulated the agent. A hidden instruction in a document, a poisoned prompt, an adversary steering the model toward an action it shouldn't take. That's a real category of risk, and it deserves the attention it's getting.

Ask AI to Choose a prompt Write a TLDR of this post Explain the security risk Summarize what CISOs should know.