AI governance is maturing as adoption grows, but employee guidance is still catching up
AI adoption is moving quickly from experimentation into everyday work, but the rules for using it are still catching up.
That shift is showing up in KPMG's Q3 2026 AI Quarterly Pulse Survey, released in September, of 314 U.S. C-suite and business leaders at companies with annual revenue of $1 billion or more. Nearly three-quarters of organizations (74%) now include cost reviews in AI approval processes, while 44% report significant workforce adoption, up from 10% a year ago.
But as AI becomes more embedded in day-to-day work, the rules around how employees should use it are still taking shape. Just 55% of organizations are developing role-specific guidelines for working with AI agents, although that figure has more than doubled from 21% a year ago.
Rahsaan Shears, a Georgia-based AI enterprise transformation leader at KPMG LLP, tied that growth to organizational maturity.
"The role-specific guidelines, this 21 to 55%, I think that speaks to maturity because you have to understand how it's integrated into the flow of work," Shears said.
As organizations move from AI pilots into wider deployment, employees need to understand where the technology fits into their work and what is expected of them. Without that clarity, adoption can be harder to translate into value, and Gallup research on AI and workplace productivity has found that employees are more likely to use and value AI when organizations provide clear direction.
Oversight is growing, but employees want answers
Companies are applying more discipline to the technology itself as AI moves from pilot projects into wider deployment. Cost reviews in AI approval processes have jumped to 74% from 61% last quarter, according to the survey, while 70% of organizations now use AI monitoring dashboards and 43% have implemented usage or token budgets, which cap how much AI processing a team can use.
Shears said that scrutiny becomes particularly important when organizations move from experimentation to scale.
"You could easily create an agent that costs more to run than it is to pay somebody to do the work if you're not careful," she said. "And when you go from pilot to scale, it becomes more and more important."
But governance isn't only about the money. KPMG found that 73% of leaders are now confident in their organizations' ability to govern AI, up from 57% last quarter, while 49% have defined high-risk use cases where autonomous decision-making by AI agents is not permitted.
That growing confidence is happening alongside wider adoption. The share of organizations building, deploying or developing AI agents has risen to 62%, according to KPMG.
For employees, though, greater availability doesn't necessarily mean greater clarity.
"There is this potential, like, am I gonna be in trouble? Is it okay?" Shears said.
Gallup's tracking of workplace AI use, which surveys employees rather than executives, found that only 25% of U.S. employees said in May 2026 that their organization had communicated a clear plan for integrating AI.
A Traliant report covered by HRD America in July 2026 found uneven AI governance maturity, with just 51% of HR teams reporting clear AI use guidelines and employee training.
Comfort, permission and performance
Shears said adoption has hit several bumps, including trust concerns and fears of job loss. But she sees three forces pushing it forward.
The first is consumer comfort. Employees are becoming accustomed to AI in their personal lives and want to do more with it at work.
"This is a time that's very different than historical technology changes where much capability in your personal life is likely more advanced than what you can do in your work life," Shears said. "And so you're learning that at home."
The second is permission.
"As more and more organizations scale, people understand what's okay and what's acceptable in their organizations for them to take advantage of," she said.
A Predictive Index report covered by HRD America found employees turning to HR and immediate peers as trusted sources of information about workplace AI.
The third is performance. Shears expects employees to lean in when leaders make AI proficiency part of what good performance looks like.
"When you hear leaders say, good performance equates to you demonstrating what you can do to get the most from AI, that'll be the carrot," she said. "It's not a stick."
Gallup's midyear analysis of AI and employee engagement similarly advised employers to spell out where employees should use AI, where they shouldn't and how managers will evaluate its use.
Who sets the rules for AI at work?
Who writes the rules varies, Shears said. The survey didn't ask which function owns the rules for how employees use AI, but she's seen the CFO, CHRO, head of IT and CEO take the lead, depending on who has the authority to establish a new organizational norm.
"Many, many times in those organizations that I think are leading the pack, you see that the CEO is at the helm," she said.
For HR, Shears sees a broader opportunity.
"The [HR] function in general really has this unique opportunity to help guide their organizations through reimagining the work," she said, pointing to redefining success and reimagining organizational structures as areas where HR can play a role.
HRD America has heard a similar argument from people leaders on why HR should lead AI implementation, particularly around governance and change management.
That work may require HR to rethink how it measures value, too.
"There are lots of things that are going to be important in the world of AI that you can't measure with many of our traditional metrics," Shears said. "That, I think, is going to be the challenge of HR organizations to understand the best and the brightest talent, how to find it, attract it, retain it, and grow it."
For companies moving AI from the pilot stage into everyday work, the technology may be changing quickly. The bigger HR challenge is figuring out how people should work with it.
