Security

Combating Shadow AI With Network Visibility

Integrated networking and enhanced observability will help states enforce DeepSeek and other AI bans.

Dan DeBacker is senior vice president at Extreme Networks.

In 2025, a growing number of states have sought to identify and block access to software or websites they deem security or data privacy threats to state devices and networks; namely, foreign-owned AI tools such as DeepSeek.

This creates a new concern for IT departments at the state and local government level: shadow AI, the unauthorized and ungoverned use of AI by employees. Those who try to use unauthorized AI tools under the radar create a host of security issues related to the movement of unsanctioned information, and the IT department lacks visibility into that.

This raises the question: How can state and local entities defend against the use of unsanctioned AI resources?

The answer? Powerful and integrated networks may be the key to providing visibility and allowing state governments to effectively monitor IT environments for banned AI tools so they can enforce policies locally.

Click the banner below to access a blueprint for better cyber resilience.

AI Bans Require Integrated Networking Platforms

With the rise of AI tools, both sanctioned and unsanctioned, state governments face increased network pressure. Additionally, cyberattackers are becoming increasingly sophisticated in their use of AI for credential theft, leading to a rise in attacks targeting state and local governments. These communities are also becoming more prominent targets for nation-state threat groups looking to influence public policy or opinion amid geopolitical uncertainty. Earlier this year, Alabama was hit by a cyberattack that disrupted state government services after some state employees’ personal information was compromised.

Despite all of this, state IT teams remain under-resourced. More than a third of government technology and security professionals report that a lack of staff makes their position very or somewhat stressful, and another third say their organization doesn’t put sufficient budget toward cybersecurity. Now, on top of all that, they have to put into place and enforce bans on popular AI tools.

State government IT teams need a way to monitor for shadow AI and cyber intrusions without adding extra work. A powerful, integrated network with built-in security and analytics capabilities is, in my experience, the best strategy to provide the visibility needed to ensure compliance and reduce IT administrator workloads.

AI itself could be used to monitor for AI. This is inevitably where we will end up in the not-too-distant future.”

One common pitfall I see in managing these networks is the separation of networking technology and administrators from the rest of the IT infrastructure. Today’s modern IT infrastructure may include multicloud environments, data centers, multiple campuses, and a multitude of devices and branches — wired, wireless, software-defined WAN and the Internet of Things, among others. One network to manage and connect all of these elements is essential.

Disparate tools create information silos, vulnerabilities and gaps in visibility. Almost 80% of CISOs and 87% of IT vice presidents report that an inability to view networking information in one location leads to challenges with security integration, according to Extreme Networks’ State of Cloud Networking 2025 report.

The only way to completely avoid complications and gaps in visibility is to combine networking, AI and security into a single integrated platform. IT leaders agree: Nearly all executives (89%, including 93% of CIOs and CISOs) would prefer to take a “platformization” approach.

Network Monitoring Can Enhance AI Observability

Network monitoring also plays a vital role in the observability of banned AI tools. AI promises to change how we do just about everything, and that’s exciting. However, not every AI tool should be used within the network. Concerns about data leakage, the use of AI for malicious intent and ensuring regulatory compliance are all valid reasons to ban tools within the IT environment. Simple application monitoring tools can be used for this; or, AI itself could be used to monitor for AI. This is inevitably where we will end up in the not-too-distant future.

DIVE DEEPER: Read CDW’s AI research report to access exclusive data.

As with many of society’s modern issues, AI is both a challenge and a solution. As a powerful monitoring tool, AI agents can identify and proactively address anomalies, such as access to certain sites and AI applications. It also reduces the burden placed on IT staff when such bans go into place, so they don’t have to manually monitor each employee’s activity.

More than a dozen states have already banned AI, and many more are considering it. State government IT teams can prepare for these bans and avoid shadow IT through modern networking technologies that provide advanced visibility and integrated analytics, keeping government systems secure.

Cecilie_Arcurs / Getty Images