Women Are Just 26% of AI Hires. That’s Not Just a Hiring Problem. It’s an AI Governance Problem.

A recent LinkedIn Economic Graph report found that AI jobs have roughly doubled in recent years, with typical salaries far above non-AI roles. Yet women account for just 26% of new hires into AI roles in the United States, and only 13% of C-suite AI leadership roles at AI companies across 27 countries.

Those numbers are usually framed as a hiring and representation problem. They are. But I think they also point to something larger: who is shaping the systems that will increasingly make decisions about everyone else?

AI is already being used in hiring, lending, healthcare, insurance, fraud detection, employee evaluation and access to services. These systems do not make decisions in a vacuum. People decide what data matters, what outcomes to optimize for, what level of error is acceptable, when a human should intervene and when a system is ready to be deployed. Those are governance decisions, not just technical ones.

There is another reason the gender imbalance should concern us. According to the International Labour Organization, female-dominated occupations are nearly twice as likely as male-dominated occupations to be exposed to transformation from generative AI. Clerical, administrative, customer relationship and business-support roles are among those particularly exposed, and many of these have traditionally provided women with stable, middle-class employment.

That creates a troubling contradiction. Women are underrepresented among the people building and leading AI at the very moment AI may disproportionately transform the kinds of jobs women have traditionally held. If a technology is going to reshape their work, their economic security and their opportunities, women should have a meaningful voice in deciding how that technology is designed, deployed and governed.

This is not simply about putting more women in the room for the sake of representation. Different experiences lead people to ask different questions. One person may see an efficient automated hiring system. Someone else may ask what happens to a candidate returning to work after years spent caregiving. One person may see a productivity gain from automating an administrative role. Someone else may ask what happens to the thousands of workers whose jobs are being redesigned around that technology, and whether they have been given any meaningful path to adapt.

My larger concern is the culture developing around AI deployment. There is enormous pressure to move quickly, automate and scale. Too often the attitude feels like deploy first and understand the consequences later. Bias testing, human oversight, appeals processes and governance can become things to figure out after the system is already operating.

That approach becomes dangerous when AI influences whether someone gets a job, receives medical care, qualifies for credit or is flagged as suspicious. By the time harm becomes visible, thousands of decisions may already have been made.

This is also why AI governance cannot belong only to engineers, data scientists or cybersecurity professionals. Their expertise is essential, but governance also requires people who understand employment, law, public policy, accessibility, communities, human behavior and institutional risk. A system can be technically secure and still produce unfair outcomes. It can be accurate on average and still fail the people who do not resemble the population around which it was designed.

The question, then, is larger than how to bring more women into AI. It is whether the institutions building and deploying AI include enough of the people who will actually live with its consequences.

Women making up just 26% of AI hires is a hiring problem. But when women are also disproportionately represented in many of the occupations AI may transform, who gets to shape that transformation becomes an AI governance problem too.

If AI is going to change people's work, opportunities and livelihoods, the people most affected should not simply be subjects of that transformation. They should have a voice in governing it.

Inspired by LinkedIn Economic Graph research on the AI talent divide, including findings that women account for 26% of new AI hires in the United States and 13% of C-suite AI leadership roles at AI companies across 27 countries.