# Datadog makes AI Guard generally available for agent security

_Published Monday, November 2, 2026 at 3:15 PM EST · Security, AI · Latest · Tier 2 — Notable_

![Datadog makes AI Guard generally available for agent security — Primary](https://web-assets.dd-static.net/42588/1793601142-ai-guard-hero-v2_resized.png)

Datadog has made AI Guard generally available, offering controls that evaluate AI agent activity as it runs and can block or sanitize prompts, model outputs, tool calls and tool results before execution. The company says the system uses conversation history, application traces and customer policies to assess actions in context rather than relying only on permissions.

Security findings reconstruct the session leading to a blocked action and connect agent activity with application and infrastructure data. Teams can block users or IP addresses and trigger remediation workflows from those findings. Real-time masking of personal information and secrets requires sensitive data scanning to be enabled.

## Sources

- [Datadog | The Monitor blog](https://www.datadoghq.com/blog/runtime-security-for-ai-agents/)

---
Canonical: https://techandbusiness.org/newswire/fE6DkqzfecBPJWxDZx0WP1
Published: 2026-11-02T20:15:58.297Z
Story chronology: 2026-11-02T00:00:00.000Z
Retrieved: 2026-11-02T22:45:06.498Z
Publisher: Tech & Business (techandbusiness.org)