Transparent by Law, Scarce by Nature: Hiring Senior AI Leaders After the EU Pay Transparency Directive

Yesterday, 7 June 2026, the deadline to transpose the EU Pay Transparency Directive (Directive (EU) 2023/970) into national law passed. Most of the coverage this week treats it as a compliance event: new reporting obligations, salary ranges in job adverts, a ban on asking candidates what they currently earn. All of that is accurate. For organisations competing for the scarcest leaders in healthcare and life sciences, such as the people who can set up a regulated clinical-AI function or lead an AI-driven drug-discovery programme, the directive also changes how those leaders are paid and which employers are best placed to hire them.

The conventional wisdom is that transparency helps the underpaid and constrains employers. The evidence is more mixed, and for senior technical hiring it points in an unexpected direction: pay transparency tends to compress the large, individually negotiated packages that scarce AI leaders have come to expect. The advantage moves away from employers who outbid others privately and towards those who can make a credible, defensible offer that others can see.

Key takeaways

• The directive’s transposition deadline passed on 7 June 2026 with only Slovakia and Italy having fully transposed. Elsewhere in the EU progress is uneven, but the content is settled: mandatory pay ranges, a ban on asking about salary history and uncapped remedies, applying to employers of every size.
• There is no executive carve-out. Recital 18 expressly covers management positions, so senior AI and data-science roles are in scope, including packages weighted towards equity.
• The law responds to a gender pay gap that has barely moved: 11.1% across the EU and 27.1% among managers. The evidence suggests the gap now comes mainly from who holds senior roles rather than from paying men and women different rates for the same job.
• Peer-reviewed economics suggests transparency reduces average wages by around 2% by weakening individual bargaining power. The effect is largest for candidates with strong bargaining positions, such as senior AI leaders, while salary-history bans raise pay for groups that were previously underpaid.
• The firms that win scarce AI leadership in healthcare and life sciences will be those that build defensible, gender-neutral pay structures now and treat a published range as an employer-brand asset, not a disclosure they resent.

Pay transparency tends to compress exactly the kind of large, individually negotiated packages that scarce AI leaders have learned to command.

Why this law exists

The directive’s purpose explains why it affects senior hiring in the way it does. It is an equal-pay measure. The principle of equal pay for equal work has been written into European law since the Treaty of Rome and now sits in Article 157 of the Treaty on the Functioning of the EU. The problem is that the principle has never fully translated into practice: women in the EU still earned, on average, 11.1% less per hour than men in 2024 (Eurostat), a figure that has narrowed only marginally in a decade. The European Commission’s assessment is that the largest part of that gap cannot be explained by measurable differences between workers, and it is not expected to close on its own, which is why the EU has turned to transparency.

What makes this directly relevant to senior hiring is where the gap concentrates. Among managers, the EU pay gap was 27.1% in 2022, more than double the figure for the economy as a whole. The higher up an organisation you look, the wider it gets. And the most useful recent evidence on European technology pay shows why: the compensation-analytics company Ravio found an unadjusted gender pay gap of 23% across European tech in 2025, falling to 2.4% when men and women in the same role, at the same level and in the same country are compared (Ravio, March 2026). Most of the gap therefore comes from who holds the senior, higher-paid roles, and what they were anchored to when they were hired, rather than about paying men and women different rates for identical work. That distinction runs through everything that follows.

It also collides with a representation problem that is sharpest in exactly the roles this article is about. Women are still only around a fifth of the EU’s ICT specialists (19.5%, Eurostat). Narrow the lens to AI specifically and the most current data; LinkedIn’s Economic Graph, reported through the World Economic Forum, puts women at 29.4% of those listing AI-engineering skills in 2025, up from 23.5% in 2018, but still under a third. The scarcity compounds with seniority: an analysis of some 1.6 million AI professionals (the think tank interface, drawing on Revelio Labs data) found women hold around 22% of AI roles globally and under 14% of senior AI executive positions. The talent pool the directive’s senior-hiring rules govern is both small and heavily male before pay is even discussed.

What actually changed

At the centre of the directive are three obligations that touch hiring directly. First, applicants are now entitled to know the initial pay or pay range for a role before they ever negotiate, typically in the job advert or before the first interview. Second, employers may no longer ask candidates about their current or historical pay. Third, and this is the part senior hiring teams routinely underestimate, “pay” is defined broadly enough to capture bonuses, equity-like instruments, allowances and occupational pensions, not just base salary. For AI leadership offers, where the headline base is often the smallest part of the package, that breadth matters enormously.

Two features make this bite at the top of the org chart. The directive contains no seniority exemption: Recital 18 expressly includes workers in management positions, and the recruitment rules apply to employers of any size, public or private. And where an employer has not met its transparency obligations, the burden of proof shifts to the employer to show that any pay difference was not discriminatory. Together these mean that undocumented, improvised executive pay is now a litigation risk as well as a fairness question, with remedies the directive leaves uncapped.

The legal map is, for now, fragmented. As the deadline passed, only Slovakia and Italy had fully transposed; major economies including Germany, France, Spain and Ireland were late, and the European Commission has confirmed it will not “stop the clock” and is prepared to open infringement proceedings against the states, reportedly at least ten, that miss it. For a multinational hiring across several EU countries, that patchwork is itself the problem: the obligations are arriving on different dates and, in several places, in stricter-than-minimum forms.

And the practical distance to travel is large. Across Europe, salary information is still the exception rather than the rule in hiring: only about 12% of German job postings disclose pay today, against 56% in the UK and 48% in the Netherlands (Indeed Hiring Lab). For most employers, then, the directive requires a real change of practice.

Why transparency can shrink the premium

This is where senior AI hiring departs from the general account. The most rigorous study of pay transparency’s effects on the market, by Zoë Cullen of Harvard and Bobak Pakzad-Hurson, published in Econometrica in 2023  , found that transparency laws lead average wages to fall by roughly 2%. The mechanism matters: once pay is visible, employers credibly refuse to pay any single individual far above their peers, because doing so triggers costly renegotiation with everyone else. The effect on wages is smallest where workers have little individual bargaining power and largest where their bargaining power is greatest. Senior AI leaders, with multiple competing offers and genuine scarcity value, have more bargaining power than almost anyone, so the large, private, one-off offer is the kind of package transparency is most likely to erode.

The picture isn’t one-directional. Salary-history bans, a core plank of the directive, work the other way for those who have historically been underpaid. Research by James Bessen and colleagues at Boston University, published in the Journal of Economic Inequality, found that after such bans employers advertised pay more often and raised offers for job-changers, with the largest gains going to women and to non-white candidates. Both findings can be true at once, and together they describe the new reality: transparency narrows gaps and lifts the floor, while compressing the negotiated premiums at the ceiling. For the kind of leaders we are discussing, it is the ceiling that moves.

This evidence comes from the United States rather than the EU, which is part of why it is useful. American states have lived with pay-transparency and salary-history rules for several years, so they are the closest thing we have to a preview of where Europe is now heading; the mechanisms are general, even where the jurisdictions differ. Read it as a strong directional signal rather than a precise forecast. But the direction is consistent and well replicated, and it is the opposite of what most boards assume transparency will do to their senior compensation.

It matters, too, that the European starting point is far lower than the US figures that dominate the headlines. The median machine-learning engineer in Germany earns around €83,000 in total compensation, against roughly $270,000 in the United States (Levels.fyi), before the frontier-lab packages that sit far above even that. Transparency operates on top of an already-compressed European base, which is part of why its effect on the bespoke premium is felt so directly here.

Why this lands hardest in healthcare and life sciences

Two things make these dynamics acute in the sectors we work in. The first is simple scarcity, and in Europe it is documented in the official statistics, not borrowed from the US. The EU had an estimated 10.3 million ICT specialists in 2024, against a Digital Decade target of 20 million by 2030: on the current trajectory, a shortfall approaching ten million (Eurostat). Of the EU enterprises that tried to recruit ICT specialists, 57.5% could not fill the roles. Cedefop, the EU’s skills agency, ranks ICT professionals as the single most shortage-prone occupation in Europe, followed, tellingly, by medical doctors and nurses. Healthcare and life sciences sit at the intersection of two of the continent’s deepest talent shortages at once. The World Economic Forum’s Future of Jobs Report 2025 tells the same story from the demand side: AI and machine-learning specialists are among the fastest-growing roles globally, and skills gaps are the single biggest barrier employers cite to transformation.

The second is that healthcare and life sciences each add their own pressure, in different ways.

In healthcare: health systems, payers, digital health, clinical-decision support, medical imaging, the binding constraint is regulation. Under the EU AI Act, AI that functions as, or within, a medical device is automatically classified as “high-risk,” bringing obligations around risk management, data governance and human oversight. That regulatory weight has created demand for a specific and rare kind of leader: one who can deploy clinical AI and govern it to a standard that regulators and clinicians will accept. The scale of the gap is visible in the numbers: a 2026 WHO/Europe survey found that 74% of EU member states already use AI-assisted diagnostics, yet only 8% have a national health-AI strategy. The technology is already in the building; the leadership to run it safely is what is missing. US health-tech firms expanding into Europe and EU health systems appointing their first Chief AI Officers are now competing for the same small pool of people who can do both.

In life sciences: pharma, biotech, drug discovery, computational biology, the constraint is competition. The organisations that need AI/ML drug-discovery leaders and principal computational biologists are bidding for them against Google DeepMind, the large technology platforms and quantitative finance. Alphabet’s Isomorphic Labs has signed major discovery alliances with Eli Lilly and Novartis, and Lilly appointed its first-ever Chief AI Officer, the computational-pathology specialist Thomas Fuchs, in late 2025. The competition is growing across Europe’s life-sciences hubs: London, Oxford and Cambridge in the UK, the biotech clusters of the Netherlands, Switzerland and Germany, and a growing AI and biology community in Paris. These employers often cannot match the cash packages offered by frontier AI laboratories, so they have to compete on the science, the mission and the credibility of the offer. Transparency increases that pressure, because it removes the option of closing the gap with a large package that no one else can see.

There is a gender dimension running underneath all of this that is easy to miss, and worth stating honestly, including where the data runs out. One might expect to be able to put a precise number on the gender gap within this specific talent pool: AI roles, in healthcare and life sciences, in Europe. One can’t. No credible source measures it cleanly, because official pay data is cut either by sector or by occupation, rarely both, and “AI” is not yet a recognised statistical occupation. That absence is itself revealing, the measurement gap mirrors the talent gap. What the evidence does show is directional and consistent. Women are markedly under-represented in computational biology and bioinformatics; typically 20–30% of researchers in the published literature. And there is a particular irony in healthcare: women make up close to 70% of the health and care workforce worldwide but only around a quarter of its senior leadership (WHO), so the very sector where women are most numerous is one where they are scarcest at the technical and executive level that AI leadership occupies. We are hiring into a pool we cannot yet fully measure, which is all the more reason to set pay openly and defensibly.

I have watched this dynamic up close. Over more than two decades placing data and AI leaders, I have seen how difficult these searches can be: handled without specialist support, they routinely run six months or longer, and prising the right person out of a major technology platform or a quantitative fund has typically meant a premium of forty to sixty per cent on their existing package. The directive does not change that scarcity. What it changes is that the premium, once paid, is now visible to the rest of the organisation, so it has to be defensible before the offer is made.

These two arenas are distinct, a point worth holding onto. The AI Act’s research exemptions mean most internal drug-discovery AI is not high-risk, so the regulatory burden that shapes clinical-AI hiring does not fall the same way on life-sciences R&D. The talent can look similar from a distance; the hiring context is not.

The advantage flips: from secret offers to credible ones

If you cannot win by outbidding in secret, how do you win? The answer the evidence points to is uncomfortable for organisations used to competing on money alone, and encouraging for those with a genuine story to tell.

Start with the fact that transparency, done well, is an attraction tool rather than a cost. A 2025 survey across six European countries found that a majority of jobseekers everywhere, from 61% in Germany to 82% in Ireland, are more likely to apply for a role when the advert shows a salary range, and that women agreed more strongly than men, which speaks directly to the directive’s purpose (Indeed Hiring Lab). The pattern holds outside Europe too, with LinkedIn’s research putting the equivalent US figure at 91%. A published range, set with conviction, expands and pre-qualifies the field of senior candidates rather than shrinking it. The organisations that treat the range as something to be dragged out of them will read, to a discerning candidate, as exactly that.

Then there is the part of the offer that transparency cannot commoditise. When the number is visible to everyone, the differentiators become the things that were always the real reasons senior people move: the significance of the problem, the quality of the team, the scientific or clinical autonomy, the conviction behind the mission. For a Chief AI Officer choosing between a health system, a biotech and a technology platform that can simply pay more, those are the levers that decide it. Transparency does not weaken that case. It makes a bidding war harder to run, which puts more weight on the case itself.

What hiring leaders should do now

Three moves are worth making regardless of where your specific EU jurisdictions stand on transposition.

First, stop asking candidates about pay history across all EU hiring today, and remove pay-secrecy clauses from contracts. This is a core obligation, it is good practice ahead of national law, and in Germany the courts are already moving in this direction: in October 2025 the Federal Labour Court ruled that a single higher-paid opposite-sex comparator can be enough to presume pay discrimination and shift the burden onto the employer. Median-only benchmarking is no longer a defence there.

Second, build a defensible, gender-neutral pay range for every senior AI and data-science role, expressed in total-compensation terms because that is how the directive defines pay, and record the reasoning at the time you set it. Germany is worth watching closely: it will miss the deadline, but its draft law is strict, with the reporting threshold expected to fall from 500 employees to 100 and a right to pay information that will apply to almost everyone. A late transposition there will not be a lenient one.

Third, treat the published range as part of your employer brand. The organisations that will struggle are those bolting transparency onto an unexamined pay structure under deadline pressure. The ones that will pull ahead are those that have already done the work to know what each role is worth, why, and how they can defend it and can therefore compete in the open with confidence.

The pay transparency era rewards a particular kind of employer: one that knows what its work is worth, can justify it, and has a reason for the right person to choose it that does not depend on a number nobody else can see. Healthcare and life-sciences organisations that build that kind of offer will be in a stronger position to hire the few leaders who decide whether their AI plans succeed.

Fergal Nolan is the founder of Banba, a specialist executive search firm for Healthcare AI and Life Sciences AI, with offices in New York, London and Berlin. He has spent more than 25 years in talent acquisition, including as a founding team member at SThree / Real Staffing and as founder of Upstream, where he has built Data Science & AI teams and AI leaders for start-ups, scale-ups and global enterprises.