AI Security and Safety: The Agent State Has Arrived | Hear what’s next from President Michael Sentonas Register now
Falcon Platform Innovations

Introducing the CrowdStrike Falcon® Platform Fall ‘26 Release

Discover the latest innovations stopping AI-powered adversaries, securing AI across the enterprise, and transforming the SOC for the AI era.

New from Fal.Con Europe 2026

Platform Innovation Highlights

AI security and safety from endpoint to cloud

Falcon Guardian unifies OS-level visibility, endpoint network inspection, agent-level controls, and an AI gateway to detect malicious activity and stop rogue agents at runtime. Secure AI agents wherever they operate, from the endpoint to cloud workloads, SaaS, and virtualized agents.

Stop rogue agents with an AI kill switch

Go beyond blocking individual prompts to stopping the agent itself. Leveraging the native Falcon sensor, the AI Kill Switch terminates malicious agents in real time and prevents them from executing again across the organization.

Achieve AI trust with agent observability

Traditional observability tools can show an agent is running, but not why it made a decision.With the planned acquisition of Fiddler, Falcon Observe will bring pre-deployment agent evaluation, continuous behavioral monitoring, and agent explainability to the Falcon platform, revealing the reasoning and decisions behind every agent action.

Find and fix AI weaknesses before attackers do

With the planned acquisition of Dreadnode, CrowdStrike will deliver automated AI security testing that runs more than 50 attack techniques, including jailbreaks and prompt injection, against models, agents, and MCP servers, with severity-ranked findings mapped to OWASP, MITRE ATLAS, and NIST AI RMF.

Detect and stop the new agent insider threat

AI agents operate with employee-level access across endpoint, cloud, and SaaS. Agent Insider Threat Detection will learn the normal behavior of each agent and its owner, flags deviations even when no policy is violated, and routes abnormal activity into multi-agent investigations for faster response.

Red team your environment with an agentic adversary

Powered by CrowdStrike SafeMind, Agentic Adversary Emulation combines offensive AI with CrowdStrike adversary intelligence to safely red team customer-approved environments, reveal exploitable attack paths, deliver prioritized remediation, and validate fixes on the Falcon platform.

Latest releases from Fal.Con Europe

AI Security & Safety
Extend enforcement with the Falcon AI Gateway

Route AI traffic through CrowdStrike’s AI gateway to inspect prompts and responses, monitor usage, and enforce policy across managed and unmanaged devices.
AI Security & Safety
Agent-level enforcement for popular AI agents

Inspect prompts and block malicious interactions in Claude Code, OpenAI Codex, GitHub Copilot, OpenCode, and Cursor, with Claude Cowork coming soon.
AI Security & Safety
Detect AI attacks with SafeMind-powered detection

Apply SafeMind frontier models and CrowdStrike adversary intelligence to detect attacks targeting AI systems and agentic misalignment.
AI Security & Safety
Identity-aware access controls for AI agents

Agentic Identity Provider integrates with AI Gateway to govern which enterprise resources each AI agent can access.
AI Security & Safety
Inventory malicious AI skills and MCP servers

Extend AI supply chain protection to discover malicious agent skills and MCP servers and stop threats before they enter your AI environment.
Proactive Security
Falcon threat intelligence, now on iOS and Android

Stay ahead of adversaries from anywhere with CrowdStrike threat intelligence in the new Falcon Threat Intelligence mobile app.
Proactive Security
Falcon Seraphic Enterprise Browser now available

Browser security is now part of the Falcon platform, stopping web threats, protecting data, and securing access without requiring a dedicated browser.
Insider Threat & Agentic SOC
Baseline every AI agent and the person behind it

Falcon Next-Gen SIEM learns normal behavior for each agent and its owner, then flags deviations even when no policy is violated.
Insider Threat & Agentic SOC
Dynamic risk scoring across identity and endpoint

Agent risk feeds the same risk score built for its owner and host across identity, endpoint, and runtime, so risky agents never sit in isolation.
Insider Threat & Agentic SOC
New detections for the AI ecosystem

Detect attacks on AI services, LLM API abuse, stolen API keys, and compromise of coding assistants, MCP servers, and agent frameworks.
Insider Threat & Agentic SOC
Search agent data where it lives

New federated search connectors for GovCloud, Dynatrace, Databricks, Snowflake, and more let you search data without re-ingestion.
Insider Threat & Agentic SOC
Multi-agent investigations now cover rogue AI agents

Expert-trained SOC agents now investigate Guardian, Cloud, Shield, Data Protection, and custom detections, including rogue and misaligned agents.
Insider Threat & Agentic SOC
Govern all SOC automation in one agentic workspace

Build, run, and govern AI agents, workflows, and custom apps in one place. Connect any agent, model, or tool and set what needs human approval.
Insider Threat & Agentic SOC
Human-in-the-loop gates for sensitive agent actions

When an agent invokes a sensitive tool, it pauses for analyst approval, with the full agent trace available for review.
Insider Threat & Agentic SOC
Pause or stop SOC agents mid-task

Analysts can interrupt a running agent to prevent unauthorized actions and runaway spend once it is already in motion.
Insider Threat & Agentic SOC
Approval gates for agents, apps, and workflows

Extend approvals beyond individual tool calls to entire agents, apps, and workflows so risky steps never run without sign-off.
Insider Threat & Agentic SOC
Real-time visibility and control over AI credits

See credits consumed by every agent and workflow in real time, halt consumption mid-run, and approve or deny limit increases.
Insider Threat & Agentic SOC
Guardrails that keep SOC agents on task

Powered by Falcon Guardian, guardrails catch policy drift, unauthorized tool calls, and out-of-bounds data access across Charlotte AI and AgentWorks.
Cloud Security
Real-time CDR now spans every major cloud

Real-Time CDR for Azure and Kubernetes expands existing AWS and Google Cloud coverage, stopping cloud breaches as they unfold.
Cloud Security
AI-based anomaly detection and posture-enriched IOAs

Surface anomalous cloud activity with AI and enrich indicators of attack with posture context to prioritize the threats that matter.
Cloud Security
Supply chain security for cloud

Extend supply chain protection to cloud environments to identify and stop compromised dependencies before they reach production.
Cloud Security
Expanded multicloud risk visibility and ASPM

Gain broader visibility into risk across multicloud environments and applications with expanded application security posture management.
Cloud Security
Expanded AI workload security

Extend Falcon Cloud Security to more AI workloads, securing the infrastructure, models, and data behind your AI applications.
Cloud Security
Protection for regulated and specialized environments

Extend Falcon Cloud Security to regulated and specialized environments with protection built for strict compliance requirements.

More from our Fal.Con US releases

Secure AI
Falcon Guardian launches end-to-end security for endpoint agents

Discover AI agents, connect their activity to endpoint execution, and stop threats before they become breaches.
Secure AI
Introducing CrowdStrike SafeMind

Purpose-built security models and harnesses from the Cyber Superintelligence Lab, built to power cyberdefense and AI safety.
Secure AI
Elite managed defense for the AI attack surface

CrowdStrike extends managed cross-domain threat hunting and MDR services to Falcon Guardian, delivering expert-led defense for autonomous AI agents.
Secure AI
Modern MDR, purpose-built to stop SaaS breaches

CrowdStrike extends elite MDR service to Falcon Shield, bringing 24/7 expert-led monitoring, investigation, and response to SaaS threats.
Secure AI
Give every AI agent a trusted identity with Agentic IdP

Continuously control agent access based on real-time risk with short-lived, tightly scoped credentials, and attribution for every action.
Secure AI
Extend modern privileged access across the AI-powered enterprise

Continuously secure privileged access across SaaS, cloud, endpoints, and enterprise resources with real-time context and centralized policy.
Secure AI
Stop supply chain attacks before code executes

Block malicious open-source packages at download across Windows, macOS, and Linux without another sensor.
Defend with AI
Multi-agent investigations across every domain

Coordinated agents investigate every domain at once with shared context. What one agent learns, they all know, converging on a single verdict.
Defend with AI
Unified, cross-domain data wherever it lives

Collect once, reuse everywhere. Falcon Onum certified pipelines clean, enrich, and route the data you bring in. Federated search reaches what you leave in place.
Defend with AI
Uncover and act on external threats faster with agentic recon

Coordinated intelligence agents continuously investigate the open, deep, and dark web to uncover and act on external threats before adversaries gain access.
Defend with AI
Orchestrate agents and automation in one workspace

One workspace for all automation. Build no-code agents with the model of your choice, develop custom apps, and connect any tool or agent through MCP.

Additional resources and news

CrowdStrike experts uncover the latest frontline intelligence. Learn how adversaries weaponize trust to build connected, cross-domain attack paths and blend into normal activity. Read the report.

CrowdStrike customers can get additional details on these and all of our releases with our platform release notes. You must have an active customer login to view release notes and documentation. Review release notes.

Many organizations think they’re prepared to face today’s ransomware, but the numbers tell a different story. CrowdStrike surveyed 1,100 global security leaders about the AI-powered tradecraft they’re seeing in the wild. Read the report.

Fal.Con brought together thousands of defenders to learn how to secure the AI revolution. Get the latest headlines from the event in our news center.

CrowdStrike recognized as a Leader

1These numbers are projected estimates of average benefit based on company’s own internal analysis and recorded metrics provided by customers during pre-sale motions that compare the value of CrowdStrike with the customer’s incumbent solution. Actual realized value will depend on the customer's module deployment and environment.

Forward-Looking Statements
This webpage may include discussion of unreleased services or features. Any unreleased services or features referenced here are still in development and subject to change. Customers should make their purchase decisions based upon features that are currently available.