Global youth unemployment hit 12.4%. What it means for an entry-level job search.

What the ILO actually reported

On 11 August 2026 the International Labour Organization published Global Employment Trends for Youth 2026: Back to the future. Its headline estimate is that the global youth unemployment rate, covering people aged 15 to 24, stood at 12.4% in 2025. That is roughly 67 million young people counted as unemployed: available for work, looking for work, and without it.

The larger number in the report is the one that gets less attention. About 257 million young people were not in employment, education, or training in 2025, roughly 20% of the world's youth population and about nine million more than two years earlier. The unemployment rate counts people who are actively searching. The NEET count includes those who have stopped.

The direction was consistent rather than isolated. Youth unemployment worsened or held flat in eight of the eleven world subregions between 2023 and 2025, and independent reporting on the release puts the split at 105 countries where the rate rose against 58 where it fell.

Job quality sits underneath the rate. In low and lower-middle-income countries, nearly nine in ten workers aged 15 to 29 are in informal employment: work without a contract, without social protection, and without the record that later counts as experience. A falling unemployment rate in those economies can coexist with almost no improvement in the security of the work itself.

The finding under the headline: the entry rungs are thinning

Most coverage of this report led with artificial intelligence. The ILO's own explanation is narrower and better supported. It points to the long decline of middle-skilled occupations, clerical and administrative work, sales roles, and production jobs. Those are the positions that historically absorbed people with limited experience and converted them into people with some. As that band of work contracts, the labour market keeps its senior rungs and its lowest-paid rungs while the step between them gets harder to reach.

The report does quantify AI, and the figure is worth reading precisely. About 6.1% of jobs held by people aged 15 to 29 are in occupations highly exposed to AI-related change. Exposure describes how much of an occupation's task content overlaps with what current systems can do. It is not a count of jobs eliminated, and it is not a forecast that those jobs will disappear.

“While the direct impact on jobs is still unclear, we must not be complacent.”

That distinction matters for anyone making a decision off this news. A shrinking band of middle-skilled work is a measured, decades-long structural change. AI displacement of entry-level roles is a hypothesis the ILO is watching and has not confirmed. Those two claims call for different responses, and only one of them is established here.

What the report does not establish

• It does not show that AI caused the rise in youth unemployment. The ILO reports exposure and says the direct job impact is still unclear.
• The 6.1% exposure share is not a count of jobs lost, and it is not a prediction that those jobs will be lost.
• Global and subregional averages do not describe any single national market, any occupation, or any employer's hiring plan.
• The 2025 values are modelled labour-market estimates that will be revised as national data arrive. They are not a final count.
• The report describes conditions through 2025. It is not a forecast of hiring in 2026 or 2027.
• Nothing in it says anything about your odds of getting an offer.

Why this lands hardest on early-career job seekers

An early-career application is usually an argument about potential, and potential is normally evidenced by a job title. The first coordinator role, the first analyst role, the first assistant role: each one exists partly to certify that someone can be trusted with the next thing. When that layer of roles contracts, the certification path contracts with it, and candidates are asked to demonstrate capability they have had no sanctioned place to demonstrate.

The work itself has not disappeared. Coursework projects, internships, shift work, campus operations, volunteer coordination, freelance tasks, and unpaid family business work all produce real, describable output. What is usually missing is a record. Six months later the specifics have gone: how many, how often, which tools, what changed. What remains is an adjective, and adjectives lose to evidence in a competitive process.

That gap is the part of this situation an individual can actually close. The rate is not yours to move. The completeness of your own record is.

What to do with this, concretely

1. List every piece of work from the last two years, paid or not: projects, shifts, internships, coursework with a real output, volunteer roles, freelance tasks.
2. For each one, write the date range, what you were responsible for, the tools or systems you used, the volume or frequency, and what measurably changed.
3. Keep numbers over adjectives. Handled roughly 40 support tickets a week beats strong communication skills in every screening process that exists.
4. Save the artifacts you are permitted to keep before access ends: your own written work, dashboards you built, schedules you ran, records you are entitled to under the terms you agreed to.
5. Write down work you were asked to do that was never in your job description. That is often the strongest evidence of scope you can offer for a role you have not formally held.
6. Revisit the list every month while the detail is still recoverable, rather than reconstructing it under time pressure during a search.

How to read the next headline about this

This report will be summarised many times over the coming months, and the summaries will drift. Three questions keep a reader oriented. Which age band is being described, 15 to 24 or 15 to 29, because the report uses both for different measures. Is the figure an unemployment rate, a NEET share, or an exposure share, because those are three different things and only the first counts people looking for work. And is the claim something the ILO measured, or something a writer inferred from it?

Applied to the coverage of this release, the question that matters most is the third one. The measured finding is a structural thinning of middle-skilled work. The inferred finding, that AI is removing entry-level jobs right now, is not something the report demonstrates. Reading news about your own labour market is a skill, and it is the same skill as reading a job posting: separate what the source states from what you are supplying.

“A generation that cannot find decent work cannot build its future with confidence.”

That is a statement about policy and institutions, and it is aimed at governments rather than at individuals. What it leaves to the individual is smaller and more immediate. A market with fewer entry rungs rewards the candidate who can show, in specific and dated terms, what they have already done. Keeping that record is not a strategy for beating a 12.4% rate. It is the ordinary preparation that makes a search shorter when it starts.