CrowdStrike Holdings, Inc. (CRWD) announced Monday that it's expanding its security platform to tackle what might sound like science fiction but is increasingly real: attackers weaponizing AI prompts to hijack enterprise systems.

The cybersecurity company rolled out Falcon AI Detection and Response (AIDR), a new offering designed to protect how organizations build, deploy, and actually use generative AI and agent-based systems. As companies rush to integrate AI into everything from development pipelines to daily workflows, CrowdStrike argues that the interaction layer where humans and AI meet has become dangerously exposed.

Protecting Every Layer of the AI Stack

With Falcon AIDR, CrowdStrike says it's delivering the industry's first unified platform that secures every enterprise AI layer. That's a comprehensive list: data, models, agents, identities, infrastructure, and interactions, covering everything from development environments to everyday workforce usage.

The company frames this as a logical extension of its core endpoint security expertise into AI systems, where prompts, decisions, and automated actions can be manipulated by adversaries who understand how these systems work.

When Prompts Become Weapons

Michael Sentonas, CrowdStrike president, explained that attackers are increasingly injecting hidden instructions into generative AI tools to influence or manipulate AI behavior. The implication is serious: if you can trick an AI agent into following malicious instructions, you've essentially compromised the system without touching traditional infrastructure.

Falcon AIDR secures prompts, responses, and agent actions in real time. By extending protection to this interaction layer, CrowdStrike aims to provide comprehensive coverage across modern AI infrastructure where systems reason, decide, and act autonomously without constant human oversight.

The company made a striking comparison: prompts now function much like traditional malware, requiring the same level of visibility and real-time defense that endpoint threats have demanded for years. It's a new attack surface that behaves like an old one.

Beyond Detection: Visibility and Compliance

Falcon AIDR isn't just about stopping threats. The platform also gives organizations deeper visibility into AI usage and agent activity, with runtime logging that enables compliance monitoring and post-incident investigations. That matters as enterprises face growing regulatory scrutiny around how they deploy and manage AI systems.

The solution integrates fully with the broader Falcon platform, allowing customers to manage AI security alongside endpoint, identity, and cloud protections through a single unified architecture. No separate dashboards or fragmented tools.

CrowdStrike shares have gained about 48% year to date, reflecting strong investor interest in cybersecurity solutions tied to AI adoption.

CRWD Price Action: CrowdStrike Holdings shares were down 1.65% at $496.45 at the time of publication on Monday.