CrowdStrike is introducing two powerful innovations in CrowdStrike Falcon® Shield to stop identity-based attacks in the AI era: a centralized view of AI agents across platforms and the integration of first-party SaaS telemetry into CrowdStrike Falcon® Next-Gen SIEM — the industry’s first native integration of SaaS security posture management (SSPM) and next-gen SIEM.
Identity has become a key battleground of modern cybersecurity. In 2024, 79% of CrowdStrike detections were malware-free as adversaries increasingly relied on credential theft, session hijacking, multifactor authentication (MFA) manipulation, and abuse of SaaS and cloud identity providers. Adversaries like SCATTERED SPIDER repeatedly demonstrate how quickly a single compromised identity can help them move laterally across domains with speed and precision. Now, as enterprises embrace more SaaS platforms, AI agents, and GenAI tools, the identity attack surface will continue to rapidly expand.
The rise of non-human identities, from service accounts to autonomous AI agents, has intensified these challenges. AI agents, for example, can write code, access sensitive repositories, invoke APIs, and trigger cross-domain workflows that directly affect production systems — yet they’re often created and deployed with minimal security oversight.
A few months ago, we launched CrowdStrike Falcon® Next-Gen Identity Security and announced new AI agent capabilities to help businesses tackle these challenges. Now, we’re building on our momentum. Falcon Shield, part of Falcon Next-Gen Identity Security, already delivers unified AI-native protection designed to stop identity-driven, cross-domain attacks — and the innovations announced today will further strengthen its ability to protect every identity as the attack surface continues to evolve.
Cross-Platform Visibility and AI Agent Governance
The proliferation of AI agents across SaaS ecosystems has introduced a new frontier of privilege and automation, and with that comes risk. Employees can build powerful AI agents that can access cloud drives, ingest corporate data, or interact with source code repositories. These agents effectively operate as autonomous and often privileged users. The problem is that most organizations have no inventory of these agents, no understanding of their access scopes, and no ability to monitor their behavior.
Falcon Shield gives them the visibility they need.
Today, Falcon Shield automatically discovers and classifies AI agents while mapping who created each agent, which users have access to it, which data it can access, what action(s) it can take, and where it’s deployed, among other information. It detects unauthorized AI agents and flags risky configurations such as internet exposure, over-permissioning, and access to code repositories. It also correlates each agent’s privileges to its associated human or service identity for full accountability.
The newest enhancements to Falcon Shield will support continuous AI agent discovery and deep visibility, which enables organizations to monitor agent behavior and mitigate risk. New innovations will include:
- Visibility of agent-to-agent activity to identify unauthorized actions or misuse
- A centralized and normalized view of agents across diverse AI platforms
- Dedicated AI agent inventory to provide continuous monitoring and investigation
- Greater insights into AI agents, including their knowledge base, specific action capabilities, and more
- The ability to alert agent owners of risky configurations or suspend agents to mitigate potential threats using CrowdStrike Falcon® Fusion SOAR