May 13, 2026
Worker Confidence Fell for the First Time in Three Years – and Organisations Are Not Tracking It
Worker confidence is falling – both in AI and in organisations’ long-term outlook. But fewer organisations are tracking employee belief, leaving leaders in the dark about how their people actually feel about the future of work.
ManpowerGroup’s 2026 Global Talent Barometer, which surveyed over 13,000 workers across 19 countries, found that overall worker confidence declined for the first time in three years. Wellbeing and job satisfaction have held steady, but confidence in the future is giving way, with AI at the centre of that decline.
While AI usage rose 13% in 2025, confidence in using the technology fell 18% over the same period. The gap between what people know they can do and what they are being asked to do next is also widening: while most workers (89%) feel confident in their existing skills, only 64% feel confident using AI and advancing technology.
Lack of psychological safety is stalling AI progress
The concern runs deeper than tools and training. New UK research from Ipsos Karian & Box finds that the barrier to AI adoption is as much psychological as it is practical. Only 38% of employees feel confident they would be supported, rather than penalised, for missing an AI error. Among junior employees, that figure drops to around 29%.
Poor training and access are not the only causes of stalled AI adoption, this data suggests. Psychological safety plays a role too, but is not getting the attention it deserves.
“There is a tendency to assume AI adoption is mainly about access to tools. What these findings suggest is that trust, clarity, and confidence may be just as important,” said Louise Breed, UK CEO of Ipsos Karian and Box.
This is not to say employees are refusing to engage with AI, or that they do not see its potential. What holds them back is scepticism around how their organisation is handling AI adoption.
While just over half of employees in Ipsos’ study believe AI has the potential to improve how their organisation works, only 39% believe their organisation is using AI to solve the right problems. The gap between belief in the potential of AI and confidence in how it is being applied sits at the centre of the story – and it is not something an AI training course can fix.
The Deeper Crisis of Organisational Confidence
This confidence decline goes beyond AI adoption. Employee confidence in the organisation as a whole is trending downwards too. New analysis from Culture Amp finds that employee belief in company success has fallen from 80% in 2021 to 72% in 2025. The data, drawn from over 3,000 companies, also found that 40% of employees are unsure or unconvinced that budgets and investments match the strategy.
Compounding this is the fact that fewer organisations are tracking employee confidence levels at all. The same analysis found that the number of organisations measuring employee belief in the company’s success has dropped by 19 percentage points over the past five years, leaving leaders with a significant blind spot about how employees view the company’s trajectory.
“Leaders are sometimes inclined to limit the questions they ask employees to only those where they expect strong results, but doing so creates blind spots,” warns Justin Anguswat, chief people and customer engagement officer at Culture Amp.
Organisations measuring the performance of their AI rollouts should take note. Examining adoption rates and training completion rates alone will paint an incomplete picture. Employee confidence in the company’s approach to AI must be measured too.
“For HR leaders, company confidence can be that practical north star: are our goals believable, are our leaders trusted, and do our systems back that up?” says Anguswat. “If the answer is no, leaders know they have some work to do. But if they’re not asking the question in the first place, how do they know where to start?”
Why Standard Responses Are Falling Short
Organisations must also not shy away from workers’ job anxieties as AI adoption intensifies. ManpowerGroup’s study found nearly half (43%) fear automation will replace them within two years.
Typical AI rollouts focus on governance, policies, training, and tools. But this approach treats people as technology users, not as human beings – and that is where the problem may lie.
“In the media, all we see is negative narrative around AI – things like deepfakes and AI-driven cybercrime,” says Erica Farmer, AI and Future Skills Specialist at Quantum Rise Talent Group. “We don’t often see the positive impact AI technologies are having on society, such as the incredible developments in healthcare and science.”
Employees are absorbing this negative coverage outside of work while being encouraged to embrace AI within it. Organisations are providing tools and training without addressing the psychological dimension, and that disconnect is creating dissonance.
Farmer believes the missing piece is making the personal case: “What organisations need to do now is engage people as individuals and tap into their ‘WIIFM’ – what’s in it for me.”
She calls this the ‘AI Dividend™’: the personal benefit an individual stands to gain from engaging with AI, professionally and in their daily life. Learning and development (L&D) teams, she argues, have a significant opportunity here. But it requires moving beyond AI literacy programmes and into territory that is harder to measure: psychological safety, validated feelings, and the time and headspace for people to genuinely explore what AI means for them.
“It takes courage to step up and make the case for the long-term approach,” says Farmer, “rather than shy away and take orders from the senior team.”
Knowing your Starting Point
What the three datasets collectively suggest is that AI adoption has become a proxy for a broader crisis of organisational trust. When employees doubt whether their company’s investments match its strategy, and when they fear being blamed for an AI error, the response cannot be limited to governance frameworks and literacy courses.
The starting point is knowing where the business stands. Organisations that are not currently measuring employee confidence levels have no baseline from which to act – and as the Culture Amp data shows, a growing number have stopped asking.
Anguswat argues that leaders should not shy away from the harder questions this kind of listening can surface. “Does your team have goal clarity? Do employees feel resources are allocated in line with these priorities, and are they being supported to achieve key outcomes?” Uncomfortable as these questions may be, the answers are what make meaningful progress possible.
