Most firms hide AI work in roles, Oyster report finds

Oyster has published research showing that 95% of companies have employees doing AI work that is not reflected in their official job descriptions. The findings point to a widening gap between day-to-day work and the way roles are formally defined.

The report combines an analysis of more than 15,000 job descriptions for workers hired through Oyster's platform, a survey of more than 1,000 HR and people-management professionals, and data from its latest impact report. It found that AI-related tasks are spreading across organisations faster than job titles, pay structures and hiring processes are being updated.

Among HR professionals surveyed, two-thirds said their companies had added AI skill requirements to roles that have not traditionally been technical, including jobs in marketing, HR and operations. Yet 46% of those employers said they had not formally communicated the change to employees doing the work.

The research also suggests that changes in responsibilities are often not matched by changes in status or pay. Among companies that added AI requirements to existing roles over the past year, 62% changed neither title nor compensation.

The pattern extends to how employers respond when staff automate part of their own work. Some 38% of HR professionals said employees would be rewarded, while 26% said nothing would change, 24% said those workers would be given more work and 6% said the role would shrink.

Most companies are trying to close AI skill gaps internally rather than through recruitment. When AI skills are missing, 53% of employers prefer to train existing employees, 31% combine training with hiring and 14% mainly hire externally.

Even where employers ask candidates about their practical AI experience, many do not yet use that information decisively. Among the 68% of companies that ask candidates what they have personally built with AI, 53% said the answer does not affect hiring or pay decisions.

Where demand sits

Data from Oyster's platform found that AI requirements have tripled since 2022, rising from 5% of roles to 15%. However, nine in 10 of those jobs do not mention AI in the title, suggesting the change is often hidden within broader role descriptions rather than clearly signalled in recruitment.

The strongest concentration of AI requirements was not in Britain. Switzerland, Poland and Portugal recorded the highest shares of roles requiring AI skills among jobs filled through Oyster, at 21%, 20% and 19% respectively, compared with 9% in the UK.

Those requirements also extend beyond software and engineering jobs. Some 17% of marketing coordinator roles, 15% of recruiter roles and 10% of operations coordinator roles filled through the platform explicitly asked for AI skills.

Hadi Moussa, Chief Executive Officer at Oyster, said the language employers use to describe work has not kept pace with change. "Every labor market runs on a shared language for describing what people can do. Titles, credentials, years on the job. AI made most of that language obsolete in a matter of months, and nobody has written the replacement yet," said Moussa.

He added: "The people making workforce decisions are often the furthest from the tools doing the work, so the shift is hardest to see from exactly the place it needs to be managed. Employers are relying on a hiring apparatus built for yesterday's work, and it's pointing them in the wrong direction. Rebuilding that judgement is the work of the next few months, and it isn't something you can automate your way into."

Looking abroad

The report found that international hiring remains a minority response to domestic AI talent shortages. Nearly eight in 10 companies said they had never turned to cross-border hiring to solve an AI skills gap at home, although 26% of those that had not done so said they were considering it.

Businesses that prioritise external hiring appear more likely to search overseas. Among employers that mainly hire externally for AI skills, 34% had hired internationally because they could not find AI-ready candidates in their home market, compared with 18% of companies focused mainly on training existing staff.

Where international hiring does take place, companies are mostly building permanent teams rather than relying solely on freelance support. Among employers that had hired internationally, 59% said they most often staffed AI-skill roles with full-time employees, 34% used a mix of employees and contractors, and 7% primarily used contractors.

The research also highlighted uncertainty over compliance when companies use cross-border contractors. Nearly three in 10 employers engaging international AI contractors said they were either not confident those arrangements were classified correctly under local labour law or had never considered the issue.

That uncertainty was higher among the heaviest users of contractors. Oyster found that 31% of companies staffing AI roles mainly with contractors lacked confidence in classification, compared with 20% of those staffing primarily with full-time employees.

Experience appears to reduce that concern. Among companies that had hired internationally for AI talent and used international contractors, 87% said they were confident those engagements were properly classified, compared with 65% of companies that had never hired internationally.

The study also pointed to limits in HR technology itself. One in five HR professionals said an AI tool or self-service HR system had missed something important in a high-stakes employee matter such as a termination or leave request.

Alongside the talent research, Oyster said its own impact data showed growing flows of work and pay across borders. The company reported USD $908.7 million in gross payment volume across 127 countries, with USD $181.5 million going to emerging markets, up from USD $57.7 million in 2022.

Moussa said international talent was often overlooked rather than unavailable. "Founders and leaders tell me every week that the talent they need doesn't exist. They do exist. We find them every day in places nobody puts on a shortlist," he said.