Shadow AI Is Outrunning the Governance Budgets Meant to Contain It
Digital Transformation

Shadow AI Is Outrunning the Governance Budgets Meant to Contain It

A survey of 400 IT and security workers found three quarters got bigger budgets this year, yet nearly half still let someone outside risk ownership decide when an AI tool connects to corporate data.

PublishedNovember 16, 2026
Read time6 min read
Share

Bigger budgets, same blind spot

A DNSFilter survey of 400 IT and cybersecurity workers found that three quarters of organizations increased their security budgets over the past year, with about one fifth seeing growth above 10 percent. Four in ten respondents said their next security dollar is earmarked for AI security and governance specifically, ahead of cloud security at 25 percent. On paper, this looks like a security function that has correctly identified its top priority and funded it.

The gap shows up when you ask what that spending has actually bought. 46 percent of respondents said they were confident unapproved AI apps were already connected to their company systems, which is itself a meaningful admission from people whose job is to prevent exactly that. One in five said an unapproved AI integration had already caused a cybersecurity incident. The budget increase and the exposure increase are happening on the same timeline, not in sequence.

Who actually decides when AI touches corporate data

The most striking figures in this report are about notification, not detection. 37 percent of organizations said they learn about employees connecting AI apps to work accounts only after the fact. 7 percent said employees are not required to notify IT about AI app connections at all. DNSFilter combines those two figures into a single number that deserves wider attention: 44 percent of organizations effectively let someone who does not own the risk make the decision to connect an AI tool to corporate data.

That is a governance design failure, not a technology gap. No amount of additional AI security budget fixes a process where the person best positioned to evaluate risk learns about a new AI integration after an employee has already granted it access. Fixing this requires a notification and approval workflow, which is organizational plumbing, not a product purchase, and it is notably absent from where the four in ten security dollars earmarked for AI governance are actually landing.

Visibility is lagging behind investment

Beyond the notification gap, the survey found 62 percent of respondents were not sure all their security tools were properly configured, and the same 62 percent could not confirm all their tools were actively monitored. Those numbers describe an environment where the tooling exists but nobody can vouch for whether it is working as intended. That uncertainty compounds specifically around AI, where new integrations appear faster than most security teams can review them.

DNSFilter's own framing is the sharpest summary of the problem: companies are failing to close critical cybersecurity gaps related to shadow AI, because the proliferation of AI has caused organizations to move faster than they might be ready for, and the policies and governance have not necessarily kept pace. That is a candid admission from a vendor whose business depends on selling the tools meant to close exactly this gap.

Why the budget to governance translation keeps failing

Budget increases are visible and easy to report to a board. Governance maturity, defined as actual enforcement, monitoring, and approval workflows functioning day to day, is much harder to demonstrate and takes longer to build. The incentive structure inside most organizations rewards the first and leaves the second chronically underfunded relative to its actual importance, which is precisely the pattern this survey captures, year over year, across one security domain after another as AI spending accelerates faster than the controls meant to govern it.

DNSFilter's phrase for this, governance spending is not the same as governance, is worth keeping on hand for the next budget review. A line item labeled AI governance that funds a policy document and a dashboard, without funding the enforcement workflow that makes employees route new AI connections through IT before granting access rather than after, has purchased the appearance of control rather than control itself. Boards reviewing that line item rarely ask the follow up question that would expose the gap: what specifically changed in how an employee connects a new AI tool to company data, as opposed to what new document now describes how that connection is supposed to work.

What a real fix requires

Closing the notification gap is the highest leverage, lowest cost fix available here, and it does not require new tooling. It requires a mandatory pre approval step for any AI integration touching corporate accounts, enforced through existing identity and access management infrastructure most enterprises already have. The barrier is rarely technical. It is that making approval mandatory slows down the exact kind of fast AI adoption that leadership has been pushing employees toward all year, and nobody wants to be the one who puts friction back into that process.

The second fix is uncomfortable in a different way: audit whether the AI security tools already purchased are actually configured and monitored, before buying more. The 62 percent uncertainty figures on configuration and monitoring suggest a meaningful share of this year's AI security budget bought shelfware, tools deployed but not verified as functioning. That audit is unglamorous work with no vendor pitching it, which is exactly why it keeps getting skipped in favor of the next purchase.

The decision for next year's budget

Every CIO and CISO setting 2027 security budgets should treat this survey as a direct challenge to justify spend differently. The question is not whether AI governance gets funded again, it almost certainly will. The question is whether that funding goes toward policy documents and dashboards that look good in a board presentation, or toward the unglamorous enforcement plumbing, mandatory pre approval, verified tool configuration, active monitoring, that actually closes the gap between spending and governing.

We think the organizations that get this right over the next budget cycle will be the ones willing to measure governance maturity by enforcement outcomes rather than by dollars allocated. Three in four budgets went up this year, and that pattern will likely repeat again next year regardless of what this survey finds. Until the notification and verification gaps this survey documents close, that spending increase is buying visibility into the scale of the problem, which is a real and useful thing to have, well before it is buying a working solution to it.

Tagged#news#digital-transformation#enterprise#cio#erp#strategy#governance#dnsfilter#shadow-ai-visibility#cybersecurity-spending#ai-tool-approval