Datadog, the monitoring and security platform for cloud applications, announced a set of new AI-powered security capabilities cloud applications, at this years Annual DASH Conference: 2025, as the industry’s first cloud-scale security platform, which introduces new tools to battle ever-evolving security threats in enterprise AI environments.
The announcements – delivered on the first day of their namesake event – highlight Datadog’s focus on helping companies safely and efficiently harness the power of AI.
One cool part of the release is Bits AI, a new platform for embedding intelligent agents that can perform vital tasks in site reliability engineering (SRE), security operations and software development.
Specific to security, Datadog launched Bits AI Security Analyst, an AI agent that is designed to independently triage Cloud SIEM signals, perform deep detections of potential threats and provide recommended responses with a rationale all without human intervention. This is intended to drive a sea change in how businesses handle security signals — mitigating alert fatigue, and speeding response times.
Datadog also unveiled the general availability of Code Security, which provides AI-powered vulnerability identification, prioritization, and remediation for custom code and open-source libraries, as well as a strong bias toward deep developer tool integration including IDEs and GitHub.
For added AI model observability, LLM Observability now tracks the model integrity and toxicity checks for prompts and responses in an organization’s AI applications.
Additionally, the Datadog Cloud Security module also offers AI security standards compliance, including the NIST AI framework, as well as misconfiguration, unpatched vulnerability and unauthorized data access detection and remediation.
The new LLM Isolation capabilities in Workload Protection (currently in preview) help identify and prevent exploitation of vulnerabilities when it comes to large language models.
Another service in preview, Sensitive Data Scanner, is designed to avoid putting sensitive information accidentally in AI training or inference data sets.
These innovative controls are just one facet of Datadog’s holistic AI security approach: proactive security monitoring across the AI stack, from development through production, and from chip to line of code, that gives organizations comprehensive visibility and automated defences against the distinct security threats of AI-native applications.
