44% of employees are already using AI at work, many without company-approved tools. And, when asked what would make them more confident about AI, reassurances around job security beat training by nearly 3:1. So what does the workforce actually need from their HR leaders when it comes to creating the culture that promotes the successful, and secure, adoption of AI?

The gap between how HR leaders and employees feel about AI is significant. 41% of employees are concerned about AI's impact on their job, but only 25% of HR leaders share that concern. This difference in perspective has potential implications for trust, communication, and how organisations approach AI adoption.

To understand how AI is affecting HR and UK workplaces generally, The Access Group, commissioned YouGov to survey 1,000 employees and 503 HR decision makers across all business sizes and sectors, asking the questions people won't raise in team meetings. Findings have also been compared with September 2024 research to track how attitudes are evolving.

Despite job concerns, the research revealed a workforce that's more ready than resistant. 70% of employees using AI report at least one tangible benefit, and among AI users, 19% are actively taking training courses while 30% are sharing tips with colleagues. But concerns remain; reliability worries have surged from 35% in 2024 to 54% in 2025, and 16% of employees are using personal AI tools like ChatGPT with company data.

Comparing September 2024 research with December 2025 reveals employees maturing in their relationship with AI. Job security concerns remain significant, but the nature of those concerns is evolving.

While job security worries persist, the 6-point drop suggests employees are slowly becoming more comfortable with AI - despite a year of alarming headlines. With 44% of employees and 73% of HR decision makers now using AI in their roles, familiarity may be reducing fear. But a new concern has overtaken it: reliability. As more people use AI daily, they're asking harder questions about whether they can trust what it tells them.

Data security concerns have also remained consistently high - 48% in 2024, 50% in 2025. Familiarity with AI hasn't eased these worries, which suggests employees need more than reassurance; they need visible evidence of how their data is being protected.

What this means for HR

This shift from abstract fears to practical questions is an opportunity. Employees aren't just asking "will AI take my job?" - they're also asking "can I trust it?" and "is my data safe?" These are questions HR can answer.
For organisations rolling out AI, this points to a clear priority: focus on building trust through transparency, clear policies, and visible safeguards - not just promoting efficiency gains.

AI adoption among HR decision makers is already widespread, with 73% now using AI in their roles. But the picture across the wider workforce is more varied.

Adoption varies by organisation size

Attitudes to AI adoption differ significantly depending on where employees work. 

Large organisation employees are most likely to use AI (47.8%), with micro businesses close behind (44.2%). But attitudes differ: only 16% of large organisation employees have "no plans" to use AI, compared to 41% in micro businesses.

This suggests that how AI is adopted, and how quickly, depends heavily on company size, available budget for AI tools, and the nature of employees' roles.

It also indicates that most employees in larger organisations are either choosing to adopt AI - with or without company approval - or are working in organisations where AI is increasingly part of company policy and technology investment.

Among employees already using AI, 40% are taking formal training to build their skills. This suggests genuine appetite for development - employees aren't just experimenting, they're investing in becoming proficient. For HR leaders, this signals an opportunity: employees are ready to learn, but may need clearer pathways and approved tools to do so safely.

Job security concerns vary significantly by sector. Understanding where employees feel most uncertain - and why - can help HR leaders tailor their approach to communication, training, and adoption.

In construction, over half of employees are concerned about job security, the highest of any sector surveyed.

This may reflect uncertainty about how AI will affect trades and site-based roles that have traditionally been less exposed to automation.

In media and marketing, concerns are almost as high (49%). Employees in these sectors are likely seeing AI content generation tools directly encroach on tasks that were previously their domain - writing, design, campaign creation - which may feel more immediately threatening than in other industries.

In IT and Telecoms, job security concerns are lowest (33%). This may reflect greater familiarity with technology, or that these employees have already integrated AI into their daily work and seen that it augments rather than replaces. It's also worth noting that AI is actively creating jobs in this sector, with growing demand for AI specialists, engineers, and implementation experts.

For HR leaders in higher-concern sectors, reassurance alone may not be enough. Employees need to see how AI will change their role, not replace it - and what skills will help them stay relevant. People won't necessarily lose their jobs to AI, but they may lose them to people who've learned to use AI effectively.

Sector spotlight: where reliability concerns are highest

Legal sector employees have the highest reliability concerns of any industry - 66% worry about AI accuracy. Finance follows at 58%.

For these sectors, the stakes around AI errors are particularly high. A hallucinated case citation or inaccurate financial figure isn't just embarrassing - it could mean regulatory breaches, legal liability, or reputational damage.

This has implications for how HR teams in regulated industries approach AI adoption. Employees may need explicit guidance on where AI can and can't be trusted, clear verification processes, and reassurance that AI outputs will be checked before action is taken. Building these safeguards visibly into workflows may help address reliability concerns while still enabling efficiency gains.

How concerns differ by age and region

AI adoption and concerns aren't evenly distributed across the workforce.

By age: Under 35s are the most active AI users - 62.5% are already using it at work, more than double the rate of over 55s (24.3%). But younger workers are also the most anxious: 50.7% of under 35s are concerned about job security, compared to 35.5% of those aged 55+.

By region: London leads on AI adoption, with 66.4% of employees using AI at work - nearly double the rate in Scotland (36.1%). But London workers are also the least concerned about job security (35.0%), while Scottish employees are the most worried (53.8%), followed by the North (46.9%).

This suggests that exposure to AI may reduce fear - but only when employees can see how it's being used and what it means for their role.

Not every AI story is about fear or frustration. With 70% of AI users reporting at least one tangible benefit, the benefits of AI in HR are real - when it's implemented thoughtfully.

What employees are gaining

• 45% report easier content generation
• 43% save time on administrative tasks
• 38% experience faster information access

What HR decision makers are gaining

Earlier we noted what HR leaders want from AI - reduced admin, faster insights, capacity for strategic work. Among those actually using AI, the research shows they're getting it:

• 50% save time on routine admin
• 48% cite faster information access
• 33% report more time for strategic work and improved decision-making

68% of HR decision makers in large organisations who use AI now work with it routinely. For these teams, AI has moved from experimentation to embedded practice - a shift that signals confidence in its value.

What employees and HR leaders are saying about the benefits of AI in HR

"Making it quicker to generate technical content." - Employee survey respondent

"Faster solutions to software problems." - Employee survey respondent

"Faster workshop/slide deck design." - Employee survey respondent

"Rewriting emails to change tone or make them more professional or concise." - HR Decision Maker survey respondent

These aren't isolated productivity hacks. They reflect RAND's recent findings that AI adoption is increasing employment, not reducing it - especially when organisations invest in new capabilities.

While the benefits are real, so are the concerns. And in many cases, employees aren't waiting for organisations to address them, they're finding their own workarounds, whether sanctioned or not.

The shadow IT problem

16% of employees are using personal AI tools like ChatGPT at work. In a 1,000-person organisation, that's 160 people potentially putting company data into unsanctioned systems - not because they're ignoring policy, but often because official approval hasn't kept pace with demand.

• 44% of employees are already using AI at work (20% company tools, 16% personal tools, 8% both)
• 26% of HR leaders use both company and personal AI tools - raising governance questions
• 50% are concerned about data security and employee privacy

For HR teams, this points to a clear priority: if employees are already adopting AI, the question isn't whether to allow it, but how to provide secure, approved alternatives before more data ends up in tools the organisation doesn't control.

AI and reliability: what it means in practice

We've already noted that reliability concerns have surged - from 35% in 2024 to 54% in 2025. But what does that look like day to day?

For many organisations, it means HR teams spending time correcting AI-generated errors, or employees losing confidence in tools that promised to save time but added a verification step instead. The disadvantages of AI in HR aren't necessarily about the technology itself, they're about what happens when implementation runs ahead of proper guardrails.

A failed AI rollout isn't just a waste of budget, it can set back adoption across the entire organisation. Being upfront about AI's limitations, rather than overselling its capabilities, may help maintain the trust needed for longer-term adoption.