CrowdStrike turns to AI as hackers get smarter

Cyberattacks aren’t what they used to be. With artificial intelligence in the mix, hackers can move faster, hide better, and hit more targets at once. CrowdStrike says its new platform can meet that challenge head-on.

At its annual Fal.Con conference in Las Vegas, the cybersecurity company announced its Agentic Security Platform, a major update to its Falcon offering. The idea is simple: fight AI with AI.

“The world is entering an arms race for AI superiority as adversaries weaponize AI to accelerate attacks,” CEO George Kurtz told the crowd. “In the AI era, security comes down to three things: the quality of your data, the speed of your response, and the precision of your enforcement.”

Washington is playing catch up

CrowdStrike isn’t the only one treating AI as an urgent threat. In Washington, agencies are moving quickly to set new rules. The Cybersecurity and Infrastructure Security Agency (CISA) has warned about poisoned data, adversarial machine learning, and deepfakes convincing enough to fool even seasoned analysts.

Those warnings are backed by data. Cisco’s 2025 Cybersecurity Readiness Index found that 86% of organizations faced AI-related security incidents in the past year, with an equal share pointing to cybersecurity talent shortages as a major hurdle.

The National Institute of Standards and Technology (NIST) has taken the lead in setting guardrails. Its Cybersecurity Framework 2.0, released in 2024, extended beyond critical infrastructure to cover all industries and reframed cybersecurity as a business risk alongside finance and reputation. NIST’s AI Risk Management Framework, together with its new generative AI profile, goes further by addressing privacy, security, and even the environmental costs of large-scale AI deployment. While voluntary, these guidelines are already shaping how companies prepare for audits and compliance reviews.

State of Cybersecurity: Healthcare’s Tactical Defense Summit – Healthcare is under siege. This isn’t theory. It’s frontline reality. Join us on Oct. 14 for a half-day virtual summit focused entirely on immediate, actionable strategies to defend against today’s most pressing cyber threats. Learn more & Register.

New tools for a new hub

At the heart of the new system is something CrowdStrike calls the Enterprise Graph, a giant data layer that pulls in information from across an organization. It acts like a constantly updating tracker of everything happening inside a company’s digital walls to give analysts a clearer view of weak points that might otherwise go unnoticed.

There’s also Charlotte AI AgentWorks, a tool that lets people build their own security “agents” using plain language prompts, no coding required. The idea is to make automation accessible to more teams, not just technical experts, to broaden the number of workers within a company who can contribute to its defense

CrowdStrike wants the latest version of its Falcon platform to serve as a hub. Using a standard called the Model Context Protocol, it connects its own agents with those built by customers or trusted partners. That structure is meant to create an environment where different tools coordinate rather than operate in silos.

Agentic AI Is transformational, but human judgment can’t be replaced

CrowdStrike’s movement towards agentic AI is part of a larger trend. Industry analysts expect AI to significantly improve security operations center efficiency, with Gartner predicting that 40% of enterprise applications will feature task-specific AI agents by 2026, transforming how security teams work.

Academic research shows AI can enhance vulnerability detection in code, with studies demonstrating AI’s ability to analyze large codebases and identify potential security flaws, though human review remains essential to avoid false positives.

But experts say there’s a catch. AI can’t replace human judgment. It’s great at crunching data and spotting patterns, but it struggles with nuance and context. That means people are still needed for strategy and decision-making.

Hackers love using AI, too

Attackers are already using AI in ways that keep security leaders up at night. Tools now scan open-source code for vulnerabilities before fixes are announced. Reinforcement learning agents can probe networks automatically, adapting their tactics in real time. Generative AI makes phishing emails harder to spot, crafting messages that mimic human writing styles or even tailoring content to specific targets. And deepfakes can mimic voices or faces with unnerving accuracy, creating new opportunities for fraud and impersonation.

These capabilities are not confined to highly resourced state actors. Off-the-shelf AI models, many of them open source, have lowered the barrier for entry. Criminal groups can now automate tasks that once required specialized expertise, such as identifying weak points in software supply chains or bypassing standard spam filters. The result is a broader field of adversaries, ranging from professional cybercrime rings to smaller opportunistic groups.

AI and people must team up as attacks evolve

The growing threats to organizations from attackers make it even more critical to balance automation with human oversight. CrowdStrike’s latest Falcon platform is pitched as a way to scale AI across an entire organization while keeping humans in control.

Most experts agree with this approach—the future of cybersecurity will likely be a mix of machines and people working together. AI brings speed and raw processing power, humans bring better judgment and context, needed when attackers change tactics or a decision has big consequences.

Keeping this balance sounds simple, but in practice it’s difficult to maintain. Security experts warn that attackers are increasingly targeting backup systems and recovery infrastructure, potentially making breach recovery more complex as organizations must secure not just primary systems but also their resilience mechanisms. The very systems that companies count on for resilience are becoming targets, too, and it’s still not clear if defenders are adapting quickly enough to stay ahead.