CrowdStrike SafeMind: When the Best Offense Builds the Best Defense

The conventional approach to AI-powered security is to build an offensive agent to find weaknesses, build a defensive agent to catch threats, and run them in separate tracks. It's a clean division of labor that produces capable tools. However, the conventional approach also produces a permanent blind spot — the defense is never trained against the actual offense, and the connective tissue between the two is manual, slow, and fragile.

With adversarial AI capable of generating and launching thousands of unique attacks at machine speed, this blind spot poses a significant and immediate risk to the enterprise.

SafeMind is CrowdStrike's answer to that gap. Developed through research conducted by the CrowdStrike Cyber Superintelligence Lab, SafeMind is a closed-loop system where offense and defense continuously sharpen each other through a process called adversarial co-evolution. The result is a defense that's been forged against the best possible attacks, run by an adversary that knows every detail of the defensive setup. This results in 70% improved accuracy, 99% cost reduction, and 6x faster detection creation that can occur autonomously at scale on new attacks.

The Architecture: Three Interlocking Pieces

SafeMind is built on three components: a high-fidelity Cyber Agent Environment that hosts an offensive agent (Red Tempest) and a defensive agent (Blue Solano) locked in continuous adversarial competition.

The Cyber Agent Environment

For adversarial co-evolution to mean anything, the agents must be orchestrated within realistic environments. CrowdStrike built the SafeMind system to deploy Cyber Agent Environments that are representative of enterprise environments. When we establish each Cyber Agent Environment, we aim to have the minimum necessary to accomplish our purpose. The system anchors on ingesting network maps and telemetry from the CrowdStrike Falcon® platform to produce a high-fidelity slice of a real environment. Each attack scenario can run more than 10,000 times to build statistically meaningful coverage.

Red Tempest

Red Tempest pursues the full attack lifecycle of network discovery, vulnerability enumeration, exploitation, privilege escalation, lateral movement, and persistence, across over 1,000 distinct attack scenarios covering 155 MITRE ATT&CK® techniques. These are long-horizon campaigns where the agent takes thousands of sequential actions across multiple consecutive runs to achieve a given objective, whether that's data exfiltration or some other form of compromise.

Under the hood, Red Tempest is a 27-billion-parameter dense model running in a multi-agent harness built around an orchestrator and swarm architecture. A manager agent tracks the overall campaign and delegates to specialized subagents — recon, assault, and others — each handling distinct phases of the operation. The offense utilizes CrowdStrike’s deep threat intelligence knowledge base containing hundreds of apex adversaries and their associated tactics, techniques, and procedures (TTPs). The context window runs at 256K tokens and can extend to 1 million. This matters because a long-running offensive campaign generates enormous amounts of state that the agent needs to reason over continuously.

Blue Solano

Blue Solano is the defensive counterpart. It operates as a custom agentic harness built around NVIDIA Nemotron Ultra as the reasoning orchestrator and a purpose-built detection generation model, post-trained on CrowdStrike's own detection engineering data.

When Red Tempest completes an attack, Blue Solano ingests the full attack trace. The orchestrator analyzes the attack description and techniques used, then queries live Falcon sensor telemetry to reconstruct what actually happened on the endpoint. Blue Solano doesn't generate detections from documentation or templates. It works against the real telemetry the attack produced, confirming what the sensor captured before writing the detection logic.

The detection generation model is a post-trained NVIDIA Nemotron Super 120B mixture-of-experts model with 12 billion active parameters. It was first trained using supervised fine-tuning on CrowdStrike's internal detection engineering corpus, encoding the equivalent of 3.1 million working hours of expertise. It was then further trained with reinforcement learning in an environment where Red Tempest attacks served as the reward signal: The model was rewarded for generating detections that caught the attack and penalized for any that produced false positives on clean data. The defense was literally trained against the offense.

On new attacks, Blue Solano calls this model to generate candidate detections, then validates them against live telemetry to confirm they catch the attack and checks them against clean baseline data to ensure they don't fire on normal activity. The output is not a report or recommendation but working detection logic. In the live SafeMind loop, Blue Solano faces attacks it has never seen before and writes the defenses for them on the spot.

The Loop That Makes It Work

Here's what makes SafeMind more than two powerful agents running in parallel.

We start by establishing the Cyber Agent Environment given to Red Tempest. From here, agent orchestration executes. Kubernetes pods with SafeMind agents deploy at scale, and the loop initiates. 

When Blue Solano successfully blocks Red Tempest, the system doesn't just log the outcome. It uses Blue Solano's new detections to harden the environment, then challenges Red Tempest to execute again with complete knowledge of every defensive change that was made. 

This is a deliberately brutal test. The cycle continues until a defined level of friction/cost is imposed on Red Tempest. At that point, the defenses have been validated against the hardest version of the adversary, one with total information.

What the Numbers Show

On the offensive side, Red Tempest achieves 100% compromise at roughly one-fifth the cost of comparable models on our internal benchmarks:

An 80% cost reduction is a decisive, game-changing factor for defense. An adversarial co-evolution loop that costs $96 per offensive cycle can't run continuously. One that costs $21 can.

Blue Solano achieves a 70% relative improvement in accuracy over general-purpose harnesses orchestrated by frontier models, while reducing the cost of generating a detection from ~$10 to $0.03. 

Performance here is measured by a strict combined metric across our internal benchmarks of previously unseen attack scenarios from Red Tempest. A detection passes only if it catches the attack and produces no significant noise when tested against clean production data. Both conditions must hold. A detection that finds the threat but floods analysts with false positives fails, just as a narrow detection that misses the attack does.

General-purpose harnesses, even when powered by frontier models, tend to cluster at higher cost and lower accuracy. The same security harness running off-the-shelf frontier models lands in the middle. Blue Solano, powered by the post-trained custom model, sits alone in the top right. The time gap tells a similar story. Blue Solano generates a validated detection in under 6 minutes. General harnesses average over an hour.

The economics of always-on autonomous defense also depend on that cost number. At $10 per detection, continuous automated defense is cost-prohibitive. At $0.03, it isn't.

Why Adversarial Co-evolution Matters

Adversaries no longer iterate on attacks manually. They are evolving to use the best off-the-shelf agents they can get in order to scale attacks today. The agent-state is shrinking the gap in capabilities between the apex adversaries and everyone else.

SafeMind is designed for the world where that's already the case. The defense loop runs continuously against the best-in-class offensive agent, improves with every cycle, and is validated against an attacker that has complete knowledge of the defensive playbook. The goal isn't to find every attack in advance, but to build defenses that hold even when the adversary knows exactly what you built.

The best security isn't the kind where you hope the adversary doesn't find your weaknesses. It's the kind where you hand them the blueprint and they still can't get through.

A First Look at SafeMind’s Functionality

The initial offering for CrowdStrike SafeMind is the Frontier AI Readiness and Resilience (FAIRR) Service, powered by Red Tempest. Red Tempest’s code review capability is one model-and-harness configuration built to systematically identify vulnerabilities in custom applications and their underlying software dependencies.

CrowdStrike will leverage Blue Solano and Red Tempest to serve as foundational components for future product offerings, while also enhancing the delivery of new and existing services.

Additional Resources

• Watch Fal.Con 2026: Future Innovation | Bartley Richardson & Alex Ionescu.
• Learn more about CrowdStrike’s Frontier AI Readiness and Resilience Service (FAIRR).
• Read about Benchmaxxing: When Benchmark Becomes the Target.
• Learn more about the CrowdStrike Cyber Superintelligence Lab.