Rising Security Budgets Fail to Contain Shadow AI Risks, DNSFilter Report Finds

Despite ramping up expenditures on defensive infrastructure, many organizations are struggling to manage the security risks posed by unsanctioned artificial intelligence deployments, according to findings from DNSFilter released this week. A substantial portion of IT and security personnel expressed doubts about the effectiveness of their organization's security governance frameworks. The research highlights a fundamental disconnect: companies are investing heavily in security tools without establishing the centralized oversight needed to manage them effectively, particularly as teams rush to acquire and deploy new defensive solutions.

Spending Up, But Governance Lags Behind

Organizations recognize the urgency of cybersecurity investments. Among the 400 IT and cybersecurity professionals surveyed by DNSFilter, three-quarters reported increasing their security budgets over the preceding year, with roughly one-fifth noting increases exceeding 10%. These allocations target traditional infrastructure including firewalls, intrusion-detection systems, identity-management platforms and related technologies.

As artificial intelligence introduces fresh security vulnerabilities while intensifying existing ones, security teams are adjusting their priorities. Forty percent of respondents indicated their next investment would target AI security and governance capabilities, substantially outpacing the second-ranked concern, cloud security, which drew 25% of responses.

Yet increased financial commitment does not guarantee network protection. DNSFilter cautioned that organizations must combine higher spending with more rigorous controls across both technical and organizational dimensions. The reality today shows many companies operating with permissive AI policies that allow security vulnerabilities to enter corporate environments undetected.

Shadow AI Creates Blind Spots

The extent of unauthorized AI tool usage reveals significant governance failures. Nearly half of IT and cybersecurity workers—46%—expressed confidence that unapproved AI applications were already connected to their company systems. One-fifth reported that an unauthorized AI integration had actually triggered a cybersecurity incident.

Discovery mechanisms are particularly weak. Thirty-seven percent of respondents said they only learn about employees connecting AI applications to work accounts after connections are already established. An additional 7% don't even require staff to notify IT departments when linking AI tools to corporate systems.

The proliferation of AI has caused organizations to move faster than they might be ready for, and while the budgets have increased, the policies and governance haven't necessarily kept pace.

DNSFilter

DNSFilter calculated that this oversight gap affects 44% of organizations, where decisions about connecting AI tools to corporate data rest with individuals who bear no responsibility for the resulting risks.

Tool Proliferation Outpaces Visibility

The challenge extends beyond AI specifically. Sixty-two percent of survey participants acknowledged uncertainty about whether all their security tools were properly configured. The same proportion could not confirm that all deployed tools were actively monitored.

Governance spending is not the same as governance. The budget bought you a policy, but the tools are still getting connected to your environment — and you can't see them.

DNSFilter