Partner POV | New hybrid mesh firewall capabilities bolster protection against AI threats

This article was written and contributed for publication by our partner, HPE.

AI is now mainstream in the enterprise, with 55% of organizations having adopted AI for business use, according to research published by the Ponemon Institute, and sponsored by HPE.1  At the same time, threat actors are leveraging the accelerating scale and increasing accuracy of generative AI to carry out even more sophisticated attacks, including persuasive, live interactions that bypass simple knowledge-based checks. This alarming issue is just one of many insights published by HPE Threat Labs in their report, In the Wild, an analysis of 1186 active threat campaigns globally from January 1 through December 31, 2025.2

What does all this mean for CISOs? The need to address AI threats is just as pressing as the need to harness its potential for transformation within the enterprise.

How can a hybrid mesh firewall help protect against AI threats?

The next evolution in firewall technology—hybrid mesh firewall (HMF)—can be a valuable tool in the security arsenal for protecting against AI threats. Hybrid mesh firewalls provide unified, multiform‑factor firewall security, giving organizations consistent policy, visibility, and enforcement across on‑premises, cloud, and remote environments. With hardware appliances, virtual firewalls, cloud‑native firewalls, and firewall as a service (FWaaS) under a single management plane, teams can apply the same rules everywhere to reduce gaps and complexity. In addition, hybrid mesh firewalls aggregate events across those environments into a single dashboard to provide a clear, contextualized view of users, data, infrastructure, and potential threat activity.

AI is everywhere, from chatbot sites to embedded applications, making it imperative for organizations to be able to deliver consistent and secure access. Hybrid mesh firewalls can provide flexible control and enforcement to manage and mitigate AI site and application risks, as well as threats from AI‑accelerated attacks, such as ransomware.

New capabilities in HPE Juniper Networking hybrid mesh firewall

At RSA Conference 2026, HPE announced new additions to the HPE Juniper Networking hybrid mesh firewall solution. These enhancements focus on giving security teams clearer visibility and policy control over AI usage, aiming to support AI governance and employee productivity while reducing the risks from sensitive data exposure and shadow IT.

The new capabilities include:

• Visibility and access management for AI websites and applications: Block access to unauthorized or high-risk AI websites, such as chatbot, search, image generation, and adult sites.
• Prompt-level inspection: Prevent data loss by inspecting and blocking submittal of prompts including sensitive keywords and/or file uploads.
• Enhanced chatbot features: Guide IT professionals in evaluating and implementing security enhancements—no specialized security expertise needed.
• Application risk assessment and control with a click: Get an at-a-glance view of applications in use within the organization, who is using them, what risk level they pose, and remediate issues with a single click.

What AI security challenges can a hybrid mesh firewall address?

Keeping up with constant change in the AI landscape is time-consuming.

Change in AI is happening at a rapid pace, with new sites, applications, and capabilities, introduced on a seemingly daily basis. For example, OpenClaw was launched in early 2026 and amassed 180,000 GitHub stars (signs of recognition) within just two weeks.3 Within the same amount of time, three high-severity vulnerabilities were identified by NIST (CVE-2026-25253, CVE-2026-25157, and CVE-2026-24763).4 In another instance of rapid change, AI chatbot provider Claude was designated a supply chain risk within days of failing to reach a negotiated agreement with the U.S. Department of War.5

New capabilities within the HPE Juniper Networking hybrid mesh firewall make supporting AI governance quicker and simpler. Security teams receive a curated list of AI sites and applications that is regularly updated, which they can accept as is or review and change if needed. Within minutes, AI site and application access policy can be updated to block even access to even newer sites, so IT teams can spend time monitoring security, not AI industry news.

Preventing data loss to AI sites and applications is difficult.

Some organizations have policies that allow employees to use AI sites and applications, but they worry about the leakage of sensitive or confidential information to the site. Users could paste confidential text, source code, or upload documents into public AI services; these inputs are often retained or used to train models, exposing IP and regulated data. For example, a user may provide confidential information in a prompt, such as, analyzing this confidential sales data and providing detailed forecast for the next 4 quarters. Or they may attempt to upload documents including private information, such as customer data.

The HPE Juniper Networking hybrid mesh firewall can inspect and block prompts based on keywords in prompt language and files. Users attempting to submit or upload documents including filtered keywords will be unable to submit their prompt to the site or application. Organizations can also tailor filtered keywords according to their own needs.

Application visibility and risk management are tedious.

Applications are the lifeblood of many organizations, yet they also represent one of the fastest growing attack surfaces. New applications often emerge, are adopted without IT oversight, and can bring various security risks to an organization, including vulnerabilities related to data controls and user privileges. For security teams, tracking and monitoring applications becomes an exercise involving multiple tools that must be correlated to get an accurate view of risk. Then, policies must be assessed, implemented, enforced, and verified across different points—a time-consuming and complex process.

HPE Juniper Networking hybrid mesh firewalls solve this issue for security teams by providing within Juniper Security Director a comprehensive at-a-glance view of applications in use within the organization, who is using them, and how risky they are. Then, with a single click, security teams can implement policy changes to reduce risk without unnecessary disruptions to operations.

 Figure 1. See applications, usage, and risk at a glance, and then take action with the click of a button in Juniper Security Director

My security experts are spending too much time on management and maintenance tasks.

Security staffing, skills, and experience shortages continue to be a top concern for organizations trying to close cybersecurity gaps. The Ponemon Institute reported in its 2026 findings that 30% of organizations cited this as a primary operational gap in their IT infrastructure, second only to difficulty in complying with IT security and privacy industry standards or regulations.6

Enhancements to the chatbot within Juniper Security Director now make it easier for any IT operator to understand security posture and review and implement change, without dedicated security expertise. For example, operators can query the bot in natural language and receive guidance on topics by asking questions such as:

• Provision the SRX for {vendor} cloud deployment
• Import policies from {vendor} firewall and translate to SRX intent with parity checks
• Identify rules with zero hits in {time window} and propose cleanup
• Create a policy that lets {role/group} access {app/service} from {locations} during {schedule} with {inspection}

With capabilities designed to address the unique risks posed by AI, organizations can more confidently accelerate security at the speed of AI.

Explore hybrid mesh firewall

How can your team benefit from a different approach to AI security? Check out these helpful resources to explore new ways to connect and protect what matters most in your organization.

References

1 , 6 "The 2026 global study on closing the IT security gap," Ponemon Institute and HPE, February 2026.

2 "HPE Threat Labs 2026—In the Wild Threat Report," HPE, February 2026.

3 "OpenClaw: What You Need to Know Before It Claws Its Way Into Your Organization," Container7, February 2026.

4 CVE-2026-25253 Detail, CVE-2026-24763 Detail, CVE-2026-25157 Detail, NIST National Vulnerability Database.

5 "Anthropic officially told by DOD that it's a supply chain risk even as Claude used in Iran," CNBC, March 2026.