Hiring used to start with an obvious problem: where do we even find this person? Post a job, pay a recruiter, wade through 300 CVs and hope the right one is somewhere in the pile.

That problem is basically solved now. AI can write the job spec, search huge talent pools, personalise outreach, screen applications and summarise interviews in a fraction of the time it used to take. Candidates are running the exact same playbook back at employers, using AI to find roles, tailor CVs, and prep for interviews. Discovery got easy on both sides at once.

What didn't get easy: actually employing the person once you've found them. That's the part 2026 is quietly exposing.

Geography Stopped Being the First Filter

If AI can match people to roles by skill rather than postcode, the sensible talent pool for any given job gets a lot bigger. Why limit a search to 30 miles of a London office when the right person might be in Lisbon, Warsaw or Manila?

Deel's 2025 State of Global Hiring report, drawn from more than a million worker contracts across 37,000+ companies in over 150 countries, backs this up with real numbers. And it pushes back on a lazy assumption:

that international hiring is mostly about chasing cheaper labour.

Deel's data on nearly 100 well-funded startups (companies founded 2020-2025 that raised $100m+) shows cross-border hiring concentrated in high-income markets, the UK led at 12.2% of hires, followed by Canada, Germany, Australia and Spain. That's not outsourcing. That's specialist skills being hired wherever they actually sit.

The clearest evidence of how fast this is moving: general AI trainer roles grew 283% in cross-border hiring in 2025 alone, the fastest-growing role on Deel's entire platform. A job category that barely existed three years ago now employs 70,000+ people across 600+ organisations, doing everything from basic data labelling to expert-level feedback in medicine and law. When a skill category can go from nonexistent to tens of thousands of cross-border hires in two years, no local recruitment market is going to keep pace with that on its own.

The CV Arms Race Is Cancelling Itself Out

There's a less flattering side to all this speed. A polished CV doesn't tell you what it used to. Candidates paste a job spec into an AI tool and get their experience rewritten around it. Employers run that CV through an AI screener. Candidate optimises for the screener, employer tunes the screener to catch the optimisation, and round it goes.

The way out isn't better CV-reading AI, it's testing what people can actually do. Give the developer a real problem. Give the marketer real data. Give the designer a real brief. AI is genuinely useful here for administering and scoring that kind of test at scale, but the point of using it should be cutting the admin around a hiring decision, not replacing the judgement in it.

Then You Say "You're Hired," and the Real Work Starts

Here's where AI-assisted discovery runs straight into a wall it can't help with.

Say the AI turns up a brilliant machine learning engineer in Brazil. Great. Now: can you legally employ them? Do you need a local entity? Are they an employee or a contractor under Brazilian law? What tax applies, what benefits are mandatory, what happens to their notice period if you ever need to let them go?

You cannot prompt-engineer your way out of employment law. More than 30 countries changed payroll, employment tax or mandatory benefit rules between 2025 and 2026 alone, and that's before accounting for the extra complexity of worker classification and employees whose work spans more than one jurisdiction.

This is the actual shape of the AI hiring story in 2026. It's not really about smarter recruiting software. It's about the infrastructure sitting underneath it, the part nobody's excited to talk about, that decides whether "we found someone brilliant" ever turns into "they actually work here."

Where Deel Fits: Underneath the AI, Not Competing With It

This is the distinction worth being precise about. You don't reach for Deel because AI helped you source someone. You reach for it because finding someone globally creates a completely different category of problem the moment you decide to hire them.

Deel provides global payroll, Employer of Record services, contractor management and compliance infrastructure across international workforces, operating in 150+ countries with entities and payroll infrastructure it owns directly in more than 130 of them. The Employer of Record model in particular changes the maths on an international hire: instead of standing up your own legal entity, negotiating local employment lawyers, and building a local payroll process from scratch, an EOR becomes the legal employer of record in that country while the person works for you day to day.

AI collapses the discovery problem. Infrastructure like Deel collapses the employment problem underneath it. Put the two together and international hiring stops being something only large companies can afford to navigate properly.

There's a Second Compliance Problem: The AI Itself

Worth being honest about this too. AI doesn't automatically make hiring fairer, it can make a biased process faster, at scale, with nobody quite able to explain why a candidate was rejected.

Regulators have noticed. In March 2026, the UK's Information Commissioner's Office published findings from its work with employers using automated decision-making in recruitment, calling for greater transparency, real safeguards, bias monitoring, and a clear route for candidates to challenge automated decisions. The EU is moving the same direction through its AI Act, which takes a risk-based approach and specifically flags employment-related AI as capable of affecting people's fundamental rights. Certain hiring-related AI systems fall under its high-risk framework, with the relevant rules currently scheduled to apply from 2 December 2027, following a change to the original implementation timetable.

"The AI said no" is not going to hold up as a compliance answer. Companies using AI in hiring will increasingly need to know exactly which systems are influencing decisions, what data trained them, and whether candidates have a genuine route to challenge an outcome.

What This Actually Means for Founders Hiring in 2026

A few practical shifts worth making rather than just nodding along to:

• Stop filtering on geography first. Start with the specific capability you need and work backwards from there, not from a 30-mile radius.

• Use AI to remove admin, not judgement. Automate scheduling, summarising and repetitive screening. Keep a human on the actual decision.

• Test skills over CVs. The easier applications are to optimise, the more valuable a real demonstration of ability becomes.

• Sort employment infrastructure before you make the offer, not after. If your candidate is overseas, know how contracts, payroll, tax and classification will work before you promise a start date.

• Treat your hiring AI as a compliance surface, not just a tool. Know what it's deciding, what it's trained on, and whether candidates have a way to push back.

The recruiting layer is getting faster and the borders are getting less relevant to discovery. None of that changes the fact that countries still have employment law, workers still need contracts, and people still need to be paid correctly and on time. That's the layer that has to work regardless of how good the AI upstream gets, and it's where platforms like Deel are increasingly doing the unglamorous, genuinely essential work.

See how Deel enables global hiring →

Quick Facts

Deel dataset size	1M+ worker contracts, 37,000+ companies, 150+ countries
Fastest-growing cross-border role (2025)	AI trainers, up 283%, now 70,000+ workers across 600+ organisations
Top cross-border hiring markets for well-funded startups	UK (12.2%), Canada (11.9%), Germany (8.8%)
Countries that changed payroll/tax/benefit rules, 2025-26	30+
EU AI Act high-risk hiring rules apply from	2 December 2027 (revised timetable)
ICO recruitment AI findings published	March 2026
Deel's owned entity/payroll footprint	130+ countries

AI Hiring Questions

Is international hiring mainly about cutting costs? Not according to Deel's 2025 data. Among well-funded startups, cross-border hiring concentrated in high-income markets like the UK, Canada and Germany, evidence that companies are hiring internationally for specific skills, not just lower wages.

What is an Employer of Record, and why does it matter for global hiring? An Employer of Record (EOR) becomes the legal employer of a worker in a country where your company doesn't have its own entity, handling local contracts, payroll, tax and compliance, while the person works for you operationally. It removes the need to set up a local company just to make one hire.

When do the EU's AI Act rules for hiring take effect? Certain hiring-related AI systems fall under the EU AI Act's high-risk framework, with the relevant obligations currently scheduled to apply from 2 December 2027, after the original timetable was revised.

Does AI in hiring need to be compliant with anything beyond data protection law? Increasingly, yes. The UK's ICO has called for transparency and bias monitoring specifically around automated hiring decisions, and the EU AI Act treats employment-related AI as high-risk given its impact on people's livelihoods, separate from standard data protection obligations.

---

This post contains affiliate links.

Really interesting take. The point about AI making it easier to find people, but not necessarily easier to actually hire them, really stood out to me.

The shift towards skills-based hiring and global talent adds another interesting layer too.

Lots to think about here...