By: Jonathan Zanger, CTO
Enterprise organizations are rapidly adopting AI tools to accelerate productivity, unlock fresh business insights, gain new competitive advantages, and drive revenue growth. But this transformation is bringing with it new and complex security challenges that introduce previously unseen risk. AI tools access sensitive enterprise data like customer records, proprietary models, internal communications, and even email content, leaving businesses without the right AI safeguards to manage new intrusion techniques and compliance risk. Employees at most businesses are using LLMs – whether their use is sanctioned or not – opening up the risk of sharing confidential data and information.
Enterprises now face the defining challenge of the remaining decade and beyond: how to harness the power of AI to produce positive business gains while guarding against new and evolving AI-driven security threats.
Rethinking Security for the AI Era
In our hyperconnected, AI-driven environment, scale and speed increasingly define both innovation and attack. Threat actors are using AI to automate reconnaissance, generate highly targeted attacks, and exploit vulnerabilities faster than traditional security operations can respond. At the same time, enterprise AI systems themselves – from copilots to autonomous agents – are introducing entirely new attack surfaces.
The only sustainable response is to rethink how security works. And to use AI to protect against AI.
That means protecting every layer of the AI ecosystem, from employees using AI tools, to applications and agents executing tasks, to the infrastructure powering the next generation of AI workloads.
At Check Point, our mission is to help customers secure their AI transformation.
AI is embedded across every application, workflow, and decision. And Check Point secures all of it.
We are already seeing the emergence of AI-native attacks targeting language interactions, hidden channels, prompt manipulation, and model-level vulnerabilities. These threats cannot be stopped by traditional security tools designed for earlier generations of computing. Preventing them requires AI-native defenses that understand how AI systems operate and how attackers attempt to exploit them.
Two important announcements at RSAC 2026 underscore how we’re leading the way in defining what AI security truly is, highlighting what we’re doing to help our customers and partners move into this new AI-driven future knowing they’re secure with Check Point by their side:
Introducing the AI Data Center Security Architecture Blueprint
Our first announcement addresses one of the fastest-growing areas of enterprise investment: private AI infrastructure.
Organizations around the world are investing hundreds of million of dollars in GPU clusters, model training environments, and inference platforms to power their AI initiatives. These environments, now commonly referred to as AI factories, have quickly become some of the most valuable assets in modern enterprises used in model training and inference. However, they are also some of the most exposed.
Unlike traditional data centers, AI environments combine high-performance GPU clusters, distributed training pipelines, massive data lakes, and real-time inference APIs. This creates attack surfaces that traditional security architectures were never designed to handle.
Our new AI Data Center Security Architecture Blueprint helps organizations solve this challenge by providing the industry’s most comprehensive security architecture for protecting AI infrastructure from GPU to gateway. We’re securing the infrastructure powering the AI revolution.
Built in collaboration with NVIDIA and leveraging Check Point’s advanced firewall and AI security technologies, the blueprint delivers defense-in-depth across four critical layers:
- Perimeter Layer
Check Point Maestro Hyperscale Firewall provides Zero Trust Network Access (ZTNA), segmentation, and scalable policy enforcement at the entry point to the AI environment - Application and LLM Layer
Check Point WAF with AI Security protects inference APIs and LLM endpoints from prompt injection, data exfiltration, adversarial queries, and API abuse - Workload and Container Layer
Integration with Illumio enables micro-segmentation and east-west traffic control within Kubernetes clusters, preventing lateral movement and isolating compromised containers - AI Hardware Layer
Through a deep partnership with NVIDIA, Check Point security capabilities are embedded directly into NVIDIA BlueField DPUs, delivering high-performance inline inspection without consuming GPU resources.
We believe AI must be secure by design, embedded across every layer of the AI stack from infrastructure to application.
Read more about our AI Datacenter Security Architecture Blueprint here.
Introducing the AI Defense Plane
Our second announcement brings the full AI security lifecycle into one unified architecture.
The AI Defense Plane is a comprehensive security framework designed to protect employees, AI applications, and autonomous agents across the enterprise.
AI has crossed an important threshold and is being used by enterprise businesses for far more than solely generating content. AI systems now connect to tools, access data, and take action on behalf of users. This agentic-driven shift creates an entirely new category of risk.
Security teams must now govern employee AI usage, monitor AI applications, and secure autonomous agents operating across cloud environments while continuously validating these systems against emerging threats.
The AI Defense Plane addresses this challenge through a coordinated platform delivering protection across three layers:
- Workforce AI Security
Visibility and governance for employee use of AI tools, copilots, and AI-enabled applications. Organizations can discover usage patterns, apply policies, and reduce the risk of data leakage while keeping user experiences seamless - AI Application and Agent Security
Following Check Point’s acquisition of Cyata, the platform expands discovery and observability for AI applications and autonomous agents—helping security teams understand what agents exist, what tools they can access, and what actions they are performing - AI Red Teaming
Now available in controlled availability, cloud-based AI Red Teaming simulates adversarial attacks against AI systems, exposing vulnerabilities in reasoning, workflows, and tool interactions before attackers can exploit them
These capabilities provide continuous visibility, guardrails, and validation across the entire AI ecosystem.
Read more about our AI Defense Plane here (link).
Securing the Future of AI Together
AI is reshaping how businesses operate, compete, and innovate. The opportunities cannot be understated. But nor can the risks.
To win in the AI era, enterprises need to build AI security into their AI transformation from the start. From agentic security operations to AI data center protection to the AI Defense Plane, Check Point is pioneering what true AI security looks like – and helping our customers and partners move confidently into this new era with knowledge and confidence.
The organizations that succeed in the AI era will not just adopt AI faster. They will embrace the right AI security approach from the start.
