Check Point launches AI Network Firewall in R82.20

Check Point has launched an AI Network Firewall as part of its R82.20 firewall software release.

The product is intended to extend AI-related security controls through firewalls customers already use, rather than requiring separate infrastructure.

The launch targets what Check Point describes as a growing gap in network security as businesses adopt generative AI tools, autonomous agents and large language models that create new forms of traffic on corporate networks. Traditional firewalls, it argues, were not built to inspect prompts, model calls or agent interactions.

Nataly Kremer, Chief Product Officer at Check Point, said the network is the main point where AI activity can be observed and governed.

"AI is transforming the enterprise network, and with the AI Network Firewall, we are transforming the firewall to secure it," said Nataly Kremer, Chief Product Officer at Check Point.

"The network is where every prompt, model call and agent interaction already converges, yet traditional firewalls were never built to see or govern that activity. By bringing AI security into the firewall organizations already run, we give security teams the visibility and control to enable AI adoption safely, in real time," Kremer said.

Blind spot

The launch reflects a wider shift in cyber security as companies try to manage employee use of public AI tools, internal AI applications, and links between software agents and external services. Check Point cited its own research showing that 87% to 93% of organisations face at least one high-risk generative AI interaction each month.

Its research also found that prompts containing sensitive corporate, personal or regulated data rose to one in every 25 interactions, double the level recorded a year earlier. Organisations now use an average of 10 AI applications a month, according to Check Point, and many are introduced outside formal approval processes.

The firewall is designed to cover three main areas: employee use of AI applications and tools, including approved and unapproved services; oversight of Model Context Protocol communications used to connect AI systems with external tools and data sources; and protection for AI applications and large language models against prompt injection and other adversarial inputs.

Check Point said it found weaknesses in 40% of the 10,000 MCP servers it reviewed. In a separate finding, its researchers identified 15,300 indirect injection payloads on public web pages, with about 70% hidden in parts of pages users do not normally see.

Existing estate

A central part of the company's case is that businesses do not want another standalone product added to already complex security estates. The AI Network Firewall runs on physical and virtual firewalls already deployed across branches, data centres, cloud environments and multi-cloud setups, according to Check Point.

Pete Finalle, Research Manager at IDC for Trusted Access and Network Security, said many businesses are wary of adding dedicated AI security products on top of existing tools.

"Organisations are challenged to deploy dedicated AI security solutions, which are additive to their existing security architecture, contributing to the sprawl of disjointed, siloed security tools and agents," said Pete Finalle, Research Manager for Trusted Access and Network Security at IDC.

"While rare, the native integration of AI security across existing enforcement points provides improvements to visibility, telemetry effectiveness, security posture and management simplicity," Finalle said.

Check Point said the firewall becomes part of its broader AI Defence Plane, which it describes as a unified control layer for AI discovery, governance and protection across networks, endpoints, cloud systems, applications and APIs. The wider architecture also covers local AI agents, software-as-a-service AI agents and controls for internally developed AI tools, it said.

Policy push

Check Point also used the launch to highlight an expansion of central policy management across its SASE and SD-WAN products. It said customers would be able to manage on-premises firewalls, cloud firewalls, AWS native firewalls, SD-WAN and SASE from one console while maintaining a single audit trail.

The platform is also intended to work across hybrid and multi-vendor environments, including integrations with IT, operational technology and micro-segmentation tools such as Illumio. The aim is to keep firewall policies aligned with changes in customer environments without manual updates.

Chris Konrad, Vice President of Global Cyber at WWT, said visibility into AI use is becoming a basic requirement for security teams.

"Policy alone will not solve shadow AI. Employees will continue using AI tools to move faster, often before security teams know those tools are present," said Chris Konrad, Vice President of Global Cyber at WWT.

"Organisations need to understand which AI applications, agents and MCP servers are interacting with their environment, and they must have the means to intercept or block unauthorized traffic. Integrating both AI visibility and active enforcement into the enterprise firewall is a practical approach because it builds on infrastructure organizations already operate and trust. That gives teams a control point to govern usage and reduce risk without slowing innovation," Konrad said.