AI Job Titles Have Left the Tech Department

Germany advertised 288 distinct AI job titles in the first quarter of 2026. Fewer than half of them sat in tech.

That single number breaks an assumption many organisations still run on. The assumption is that AI work belongs to the technology function, and that governing it is therefore a technology problem. The data says otherwise. AI job titles have spread into sales, human resources, legal and administration. The compliance duties that attach to AI systems have quietly followed them out of the server room.

AI job titles left the tech department

Indeed’s Hiring Lab tracked AI-labelled job titles across five European labour markets. It found sharp growth everywhere it looked. Germany went from 72 such titles in 2022 to 288 in the first quarter of 2026, which is 4.2% of all job titles advertised, up from 0.8% four years earlier. France climbed from 35 to 138, or 3.3%. The United Kingdom moved from 61 to 160, at 2.7%. Spain rose from 8 to 81, and the Netherlands from 21 to 84.

Growth alone is not the story, though. The interesting figure is where those roles sit.

In Germany, 59% of AI job titles are now outside tech occupations. The Netherlands sits at 58%, while France and the United Kingdom both land at 54%. Only Spain runs the other way, with 64% of its AI-labelled roles still inside tech.

So in four of the five markets, the typical person hired under one of these AI job titles is not an engineer. She is a marketer, a recruiter, a paralegal or an operations analyst whose job description now carries an AI qualifier.

Why the AI job titles signal matters

A job title is a weak signal on its own. Employers inflate them, and a title containing the word “AI” does not prove the role touches a regulated system. Taken across five countries and four years, however, the pattern is harder to dismiss.

Employers are not merely buying AI tools and dropping them into existing workflows. They are rewriting the roles themselves. That is a much stronger commitment, and a much slower one to reverse.

HSBC hired governance, not only engineers

A useful illustration landed on 27 July. HSBC announced a Global AI Centre of Excellence in Singapore, opening in the second half of 2026, with more than 100 roles attached to it. The interesting part is the composition. Alongside natural language processing and data science, the bank listed AI governance and human-centred design as capabilities it intends to staff.

Read the bank’s AI job titles against the Indeed figures and a shape emerges. A regulated bank is treating governance as a hiring line rather than a policy document. That is what happens once AI moves from pilot to production and someone has to answer for it.

The centre’s first projects point the same way: customer wealth conversations, agentic treasury operations and AI-enabled payments. None of those is a laboratory exercise. Each one touches a customer, a transaction or a regulated decision. Each therefore needs a named person who can explain what the system did.

AI job titles carry legal duties with them

Here is where the labour-market data stops being an HR curiosity. Two provisions of the AI Act attach obligations to people rather than to systems. Both of them now reach well beyond the technology function.

Article 4 got lighter, not narrower

The AI literacy duty has applied since 2 February 2025. In its original form it required providers and deployers to ensure “a sufficient level of AI literacy” among staff and others operating AI systems on their behalf.

That standard did not survive the week. The Digital Omnibus entered into force on 27 July and softened it. How far is contested: the Commission’s own summary describes the company-level requirement as replaced by non-binding encouragement, while legal commentators read the amended text as a duty that survives in weaker form, obliging organisations to support and promote literacy rather than guarantee a level. That gap is worth watching rather than resolving today.

Either reading points the same way, though. The obligation got lighter exactly as the population it covers got wider. Nothing in the amendment confines literacy to technical staff, and nothing carves out a recruiter using a screening tool or a claims handler working an automated triage queue. Where the work migrated, the expectation migrated too. What changed is how hard anyone can be made to meet it.

Article 26 needs competence, training and authority

For deployers of high-risk systems the requirement is sharper still. Article 26(2) obliges them to assign human oversight to natural persons who have “the necessary competence, training and authority, as well as the necessary support”.

Authority is the word that tends to get skipped. Competence and training can be bought through a course. Authority cannot, because it is a question of whether the person watching the system is allowed to stop it. If AI job titles now sit in HR and legal and customer operations, then oversight authority has to sit there too. That is an org-chart decision rather than a training decision.

What AI job titles mean for the org chart

The gap worth auditing is not skills. It is the distance between where AI systems are operated and where the accountability for them is currently written down.

Four questions make that gap visible:

• Which roles outside the technology function now operate, configure or act on the output of an AI system, whatever their title says
• Whether those people fall inside the AI literacy programme, or were scoped out of it as non-technical
• Who holds documented authority to pause or override each system, and whether that person sits close enough to the work to notice a problem
• Whether job descriptions rewritten to include AI tasks were also reviewed for the duties those tasks carry

Plenty of AI governance programmes were designed when AI lived in one department. The Indeed data suggests that assumption expired some years ago, and the roles have kept moving since.

The quiet reorganisation

There is a version of this story that reads as good news, and it is largely true. AI capability spreading across a business is what adoption looks like when it works, whatever it may cost the junior pipeline later. A bank staffing governance roles is a sign of maturity rather than panic.

But diffusion carries a governance cost that arrives later than the capability does.

When AI sat in a single team, oversight was a small problem with a clear owner. Now that AI job titles are distributed across most functions, oversight becomes a coordination problem. Coordination problems fail quietly. Nobody announces that no one was watching. It surfaces in an incident review months afterwards, when someone asks who was supposed to catch it.

The organisations that handle this well will not be the ones with the largest AI teams. They will be the ones that noticed AI job titles had left the building, and moved the accountability out after them.

If you want to check where your own gap sits, the Future Prep free assessments and checklists are built for exactly this kind of stocktake.