A comprehensive VentureBeat Pulse Research survey of 145 senior IT executives at companies with over 100 employees reveals a widening gap between aggressive AI adoption and the ability to govern it. The report shows that nearly three in five enterprises (58%) are expanding their AI initiatives, but only one in ten (10%) have active automated monitoring in production. The core issue is not technological but organizational: a missing single accountable owner for AI across the stack.
58% of enterprises accelerate AI while 23% rationalize
Over the past twelve months, most companies have grown their AI portfolios. 33% report significant expansion, while 25% see net positive growth. However, a quarter of organizations (23%) are actively rationalizing, cutting what doesn't work and focusing on what delivers. Another 12% keep portfolios steady. Only 3% have paused expansion to first address governance. This innovation race creates the control gap: companies accelerate without the visibility needed to manage risks.
Sponsored Protocol
85% of enterprises run two or more platforms each claiming primacy
One of the most striking findings is the fragmentation of the AI landscape. 85% of enterprises use two or more platforms, each positioning itself as the primary AI layer. Whether ERP, EHR, ITSM, productivity suites, or data platforms, every system claims a central role. Only 8% have consolidated to a single layer. This makes cross-platform governance extremely complex, as there is no shared center of gravity.
Only 10% have active monitoring, 30% rely on manual review
At the heart of the problem is detection. 40% of executives are very confident they could detect a model drifting, behaving unsafely, or failing in production. However, the vast majority of this confidence rests on manual human review (30%), while only 10% have active monitoring and alerting in place. At the other extreme, 8% have no systematic visibility, and 19% would only learn of issues from end users. In practice, over a quarter of companies would discover a failure only after it has caused damage.
Sponsored Protocol
49% suffer from shadow AI, 25% faced runaway agent bills
The financial and operational consequences are already tangible. Nearly half of enterprises (49%) cite shadow AI as their most severe failure — unauthorized pipelines run on corporate cards outside central oversight. Another 25% have been hit by runaway infinite-loop agent bills, and 6% saw agents degrade production databases. Only 21% report having imposed hard token and budget caps at the infrastructure level, avoiding surprises. In short, four in five companies have already experienced a concrete AI-related control failure.
The biggest barrier is the missing accountable owner
When asked about the single biggest barrier to governing AI across platforms, 32% point to the absence of a single accountable owner. 25% cite vendor opacity, 16% lack of observability tools, and 17% leadership deprioritization. Only 5% cite talent shortage. Even where a central governance team exists (38%), the accountable role is often the CTO/CIO (27%) or Chief AI Officer (22%), but in 17% of cases no one holds formal accountability.
Sponsored Protocol
Interestingly, even consumer platforms like WhatsApp introduce AI features with impersonation risks, highlighting how governance must extend to all levels. According to Wikipedia on AI governance, lack of transparency is a recurring theme.
In conclusion, the control gap is not a technology problem solvable by more spending, but an ownership and accountability issue. Enterprises that want to fully leverage AI must first establish a central authority capable of seeing, controlling, and governing the entire stack; otherwise, failures will continue to multiply.
