CrowdStrike launches AI security tools for enterprises

CrowdStrike has unveiled new security products for AI-driven operations, software supply chain protection and identity management, expanding its Falcon platform into three areas of growing concern for corporate security teams.

The most prominent update is what CrowdStrike describes as the next stage of its Agentic Security Operations Centre. Multiple AI agents can investigate a security incident across endpoint, identity, software-as-a-service, cloud and network environments at the same time.

The system is designed to address attacks that move across different systems in parallel rather than along a linear path. In that model, separate tools focused on separate domains can leave analysts to piece together incomplete findings after the event.

CrowdStrike said its approach allows domain-specific agents to work from a shared context layer, enabling each one to use information gathered by the others during the same investigation. The aim is to produce a single conclusion rather than a string of disconnected alerts.

The updated system also covers threats targeting enterprise AI systems, including model abuse, prompt injection and data exfiltration through AI assistants. Alongside the investigation tools, CrowdStrike introduced a single workspace for building and governing automated response processes with varying levels of human oversight.

Michael Sentonas, President, CrowdStrike, said the changes reflect the pace and spread of AI-based attacks. "We're already seeing AI agents execute attacks across multiple systems at the same time.

"AI agents in the SOC are table stakes. Agents working together across every domain, on the same investigation, that's the new standard. CrowdStrike's architecture and expert validation make this possible, and confidently answer the question every CISO is asking: how do I trust what my agents found, and how do I know it's right?"

Supply chain

A second product launch focused on software supply chain attacks, an area that has become more urgent as developers and automated coding tools rely heavily on open-source packages from public repositories.

CrowdStrike introduced Real-Time Supply Chain Attack Protection, which it said blocks malicious open-source packages at the endpoint before any embedded code can run. The tool targets package manager activity such as npm and pip installations across Windows, macOS and Linux.

CrowdStrike argues that the endpoint has become the critical control point because harmful package code executes at installation. It said the product can intercept those transactions before scripts run, then trigger investigation and remediation across affected endpoints.

The product also gives security teams a package inventory across endpoints, allowing them to identify where a compromised dependency has been installed. Teams can set controls such as minimum package age requirements in an effort to reduce exposure to newly poisoned packages.

CrowdStrike linked the launch to threat activity it has observed in package ecosystems. It cited findings from its threat-hunting research that said a DPRK-linked actor known as STARDUST CHOLLIMA poisoned 131 trusted AI framework packages, while an eCrime actor called ALTERED SPIDER compromised more than 300 software dependencies in a single day.

Sentonas said supply chain compromise had become a major entry point for attackers. "Attackers know that compromising one trusted package can give them a path into thousands of organisations. That makes the software supply chain one of the most powerful attack surfaces in the AI era.

"The endpoint is where malicious code executes, and only CrowdStrike turns it into the control point that stops the attack."

Identity controls

The third launch was an Agentic Identity Provider, intended to create and manage trusted identities for AI agents acting inside corporate systems.

The move addresses a problem now emerging as software agents carry out tasks, access applications and move data with little direct human involvement. Traditional identity systems were built around user logins, passwords and manually provisioned accounts, whereas autonomous agents are often represented through service accounts, API keys or workload identities.

CrowdStrike said its Agentic Identity Provider automatically registers AI agents when they come online, issues cryptographically verifiable identities and grants short-lived access tokens limited to the minimum access needed for each task. It also said every action can be tied back to the human user or workload on whose behalf the agent is acting.

The product sits alongside the company's wider Continuous Identity model, which replaces standing access rights with real-time authorisation and revocation. CrowdStrike argues that before an AI agent can be continuously authorised, it first has to be established as a recognised identity within the organisation's control systems.

Scott Kriz, GM of Continuous Identity at CrowdStrike, said existing identity products were not designed for autonomous software actors. "Securing AI agents demands solutions built for how they operate.

"Continuous Identity modernised identity security for the agentic era, but you cannot continuously authorise an identity you were never able to establish, and traditional identity providers break the moment an agent acts on its own. Agentic IdP is the identity provider for AI agents."