A survey of IT professionals found widespread AI use, operationally risky hallucinations and gaps in governance ownership.
AI use brings operational risks into view
The Fast Company survey covered one thousand five hundred IT professionals. Sixty-eight per cent said they had seen AI create a hallucination with potential operational impact. That finding records exposure to an output that could affect operations; it does not mean every respondent experienced a resulting failure. The distinction matters because the survey separately asked whether hallucinations had caused issues. Together, the responses show why AI use and operational risk need to be considered side by side: the first measure captures what professionals have witnessed, while the second counts reported consequences.¹
Sixteen per cent said hallucinations caused issues, while two per cent reported not using AI at all. These figures measure different things: one captures reported consequences, and the other shows how few respondents were outside AI use. They do not establish how often a system hallucinates, which teams use it or how severe the reported issues were. They do show that the operational question reaches a workforce where AI use is widespread. Governance decisions therefore concern more than the cases in which someone has already reported a problem.¹
Named owners do not guarantee clear accountability Eighty-five per cent of IT professionals said every AI agent and workflow in their organization had a named owner.
Named owners do not guarantee clear accountability
Future Skills & WorkDecentralized Labor Governance Is Redrawing the Map of Work
Meanwhile, 40% of gig workers now operate outside formal labor codes, embedding themselves in decentralized networks where algorithmic sovereignty.
Read More →Eighty-five per cent of IT professionals said every AI agent and workflow in their organization had a named owner. That is a formal assignment of responsibility: each system or process has someone attached to it. The survey figure does not establish what authority that person has, what checks they perform or how they respond when an output causes a problem. A named owner can make responsibility traceable in principle, but the existence of a name alone does not show that people know who can act. The gap between assignment and practical accountability is central to the governance findings.¹
Only forty-two per cent said ownership was clear in practice. The contrast points to a difference between recording an owner and making responsibility understood across an organization. Employees also gave a low rating to policy follow-through: just twenty-four per cent said governance policies were followed very consistently. These measures do not explain why ownership or compliance falls short. They do show that written assignments and rules do not, by themselves, establish consistent practice. The survey presents clarity and adherence as separate weaknesses in turning AI governance into routine organizational behaviour.¹
Governance is a reported deployment barrier
Twenty-seven per cent of IT professionals named governance, security or compliance as the top barrier slowing AI deployment. The combined category ranked ahead of skills shortages and technology limits. That makes governance a practical constraint on putting AI to work, as well as a response to operational risk. The finding does not identify which rules, security concerns or compliance requirements create the slowdown, or how long deployment takes. It does place governance alongside the capabilities and technology needed to deploy AI: respondents cited it as the leading barrier among those options.¹
The survey also links advanced AI use with a concrete time saving: IT professionals using AI at the most advanced level said they saved an average of six hours a week. That figure describes the reported average for this group; it does not say that every user saves the same amount or that the saving applies across all organizations. Read alongside the reported deployment barrier and gaps in governance practice, it shows two sides of AI use in the survey: advanced users report time saved, while organizations still cite governance, security or compliance as an obstacle to broader deployment.¹
Ask Who Can Act, Not Just Who Owns
When you evaluate an AI role or a project using AI, do not accept a named owner as proof of safety. The survey shows eighty-five per cent of organizations have a named owner for each AI agent, yet only forty-two per cent say ownership is clear in practice. In interviews, ask who has authority to stop a flawed output and what happens when one appears. A title on paper does not mean anyone can act.¹
Future Skills & WorkMore Than 138,000 NYC Students Were Homeless Last School Year
The decline came as families left shelters for subsidized housing; advocates also cited migrant families avoiding authorities by leaving shelters and school.
Read More →Governance is a reported deployment barrier Twenty-seven per cent of IT professionals named governance, security or compliance as the top barrier slowing AI deployment.
If you build AI skills, the payoff is real: advanced users report saving six hours a week. But plan for governance friction as a career risk. Twenty-seven per cent of IT professionals name governance, security, or compliance as the top barrier to deployment, ahead of skills shortages. That means your next AI project may stall on policy, not ability. Budget time for approvals and clarify escalation paths before you commit to a deadline.¹
Sources
Written by Career Ahead AI. AI can make mistakes, so verify key facts. Report errors to support@careeraheadonline.com
