In brief: Check Point is adding an AI Network Firewall to the R82.20 firewall software release that detects AI usage, MCP communication, and prompt injections at existing network control points.
With the R82.20 firewall software release, Check Point has introduced an AI Network Firewall that detects and controls AI usage, MCP communication, and prompt injections at existing control points. This allows AI security to be integrated into existing network architectures without additional infrastructure.
Check Point is extending the R82.20 firewall software release with a feature called AI Network Firewall. It is designed to identify and control AI-related traffic directly at a network’s existing control points – without requiring additional hardware or separate security layers. Three areas are intended to be detected: the general use of AI services on the network, communication via the Model Context Protocol (MCP), and prompt injection attacks, in which inputs are manipulated to induce undesired behavior in AI models.
For CISOs, this creates a starting point for making the growing use of AI applications and agents in the corporate network visible and regulating it, without fundamentally overhauling the existing security architecture. MCP detection in particular is relevant, as this protocol is increasingly used to connect AI agents to internal systems and data sources, thereby opening up new attack surfaces that traditional firewalls have not previously addressed. Prompt injection is considered one of the most practically relevant attack techniques against AI systems in production use, as it directly targets weaknesses in natural language processing.
For an assessment in one’s own environment, it is relevant that the feature is being integrated into an existing product line and should therefore be usable by Check Point’s existing customers with reasonable effort. The announcement does not specify the technical details of how prompt injection detection is performed or what detection rate can be expected. Before rolling this out, CISOs should check which AI services and MCP connections are already in use on their own network in order to configure the new feature accordingly.
Source: www.security-insider.de · Published August 19, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.8.3.