A survey of 810 public sector information technology (IT) practitioners and leaders found a wide gap between how fast governments were adopting AI and how mature their governance and monitoring controls were, according to a MeriTalk report.

According to the research report released on September 22, only 37% of the respondents have a formal AI governance framework that’s actively enforced. Twenty-eight percent said their framework is incomplete or inconsistently applied, and 35% said they have no framework at all.

The survey was conducted across the United States and United Kingdom in June by research firm UserEvidence on behalf of SolarWinds, an observability and IT management software provider.

“Confidence in AI governance is running well ahead of the actual controls agencies have in place, and that gap is where risk quietly accumulates,” Travis Galloway, senior advisor for government affairs at SolarWinds, said in a press release announcing the findings.

State and Local Governments Lagging Behind Federal Governments

The survey found a sharp divide between federal respondents and their state and local counterparts. 

Forty-three percent of federal respondents said they have a formal governance framework that’s actively enforced, compared with just 19% at the state and local level, according to the report.

Federal respondents also reported moving faster on AI overall. Thirty-two percent described their organization as aggressively pursuing AI adoption, compared with 12% of state and local respondents, the report found.

Security Incidents Are Rising Faster Than Monitoring Can Keep Up

Even with more mature governance, fifty-eight percent of federal respondents said AI-related security incidents increased over the past 12 months, compared with 37% at the state and local level, according to the report.

Monitoring was also found to be lagging. 

While 80% of respondents across the full sample described their overall IT monitoring as comprehensive or mostly comprehensive, only 38% said AI tool behavior is fully built into that standard monitoring. Forty-seven percent said they monitor AI tools only partially, and 15% said they don’t monitor AI tool behavior at all, the report found.

Staffing is a shared problem across both levels of government. 

Forty-five percent of federal respondents and 43% of state and local respondents both cited a lack of skilled staff as a top barrier to AI adoption, according to the report. State and local respondents were far more likely to cite budget constraints (40%), compared to just 24% of federal respondents.

A Core Issue In AI Governance

For chief data officers (CDOs), this survey reflects a familiar pattern: confidence in AI governance is running well ahead of the actual controls agencies have in place, and that gap is the real risk.

The monitoring numbers are the most actionable part of this survey. If only 38% of organizations have fully integrated AI tool behavior into their standard monitoring, that means most organizations still can’t clearly answer basic questions about what their AI systems are doing. 

The core issue in AI governance is that teams need to be able to see what AI systems are doing before they can reliably enforce rules about their behavior. According to this survey, most organizations lack both: visibility (they don’t fully monitor AI behavior), and enforcement (fully enforced governance frameworks).