The world: rich countries first

The jobs most exposed to AI are in rich countries: 34 percent of jobs in high-income countries, against 11 percent in low-income ones, by the UN's labour agency's count.

The countries that grew on the rich world's desk work are losing their next rung: the Philippines' outsourcing industry now expects between 360,000 and 650,000 fewer call-centre and back-office jobs by 2028 than it had planned.

Whether a worker is protected depends on where they live: a court in China, a law in Europe and one state in America each answer differently.

When factories automated, the jobs that went first were in the places that made things cheaply. When work was outsourced, it moved from rich countries to poorer ones. Technology shocks usually land hardest on the people with the least. Most people assume AI will do the same.

So we went looking past America. At the global estimates of which jobs AI can reach. At the countries that do the rich world's office work. At the young abroad. And at what courts and governments are doing about it.

Here is what we found. Generative AI runs the other way. It works on text, numbers and code, the desk work that rich economies are built on. So the richer the country, the more of its work is exposed. The countries that grew by doing that work are already cutting their own forecasts. And the protection a worker gets depends on the country they work in.

The International Labour Organization, the UN's agency for work, has the best global estimates. "Exposed" here means that much of a job's work is the kind generative AI can do or change. It does not mean the job will be lost.

In May 2025 the ILO found about one job in four worldwide is potentially exposed. In high-income countries the figure is 34 percent. In low-income countries it is 11 percent.

The joint study also found the other side of the coin. Of the 441.8 million jobs in poorer countries that AI could help, 66.9 million belong to workers with no internet connection to use it. The poor are less exposed, and also less able to gain.

The exposure falls unevenly inside rich countries too. Women's jobs there are about twice as likely as men's to be in the most exposed group, 8.5 percent against 3.9 percent. The reason is plain. So many clerical and administrative jobs are held by women.

Between the rich world and the poor sits a group of economies that grew by doing the rich world's office work: answering its calls, writing its code, processing its paperwork. Their whole model rests on the work AI is best at.

The Philippines. The country's call-centre and back-office industry employed about 1.9 million people in 2025, by its industry body's own count. That body is IBPAP. In July 2026 it cut its forecast. It had expected 2.5 million workers by 2028. It now expects 2.14 million at best and 1.85 million at worst, fewer than it has today. Between about 360,000 and 650,000 jobs that were going to exist will not.

Nobody is laid off in that sentence. Its president and chief executive, Jack Madrid, credits AI. He also credits buyers' behaviour, geopolitics and rival hubs from Egypt to Colombia. He does not call it an AI-only story, which is why it is credible.

India. The largest IT services firms are the backbone of India's middle class. Tata Consultancy Services, the biggest, announced about 12,200 cuts in July 2025, about 2 percent of its staff, mostly in middle and senior management. It called them part of a future-ready transformation. It did not blame AI.

By April 2026, the top firms' combined headcount had fallen by 7,389 in the fiscal year, as the Indian news agency IANS counted it. Another tally puts the fall at 6,981. The industry body, Nasscom, says the sector's headcount is still growing, by about 2.3 percent. That is the industry's own figure.

Ireland. Ireland hosts the European headquarters of the largest tech companies. In the year to early 2026, its information and communication jobs fell by 20,300, about a tenth, mostly in programming. One caveat matters. The statistics office changed how it classifies industries that same quarter, so the exact size is uncertain. Ireland's own research institute estimates about 7 percent of Irish jobs could be displaced, with tax receipts falling and welfare costs rising.

A documentary from CNA in Singapore shows workers in India and the Philippines training the AI systems that are meant to replace them.

The bottom rung is going missing outside America too.

In Britain, graduate vacancies fell to their lowest level since 2016 by July 2026, down 45.6 percent. The big accounting firms cut graduate intakes and said openly that AI was taking junior work. In October, Al Jazeera asked why Britain is at the forefront of AI-driven job cuts.

In South Korea, the Bank of Korea found that of the jobs lost among young workers, 268,000, or 94 percent, were in sectors highly exposed to AI.

The head of the International Monetary Fund warned in January 2026 of AI "like a tsunami hitting the labor market" for young people.

Countries are answering in very different ways.

China's courts. In 2026, the Hangzhou Intermediate People's Court found it unlawful to dismiss an AI content reviewer. The worker had refused a transfer carrying a 40 percent pay cut after AI took over the work. We know it as one of a batch of cases the Hangzhou courts chose to publicise. No judgment number or date has been published, and the worker is not named.

Europe. The EU's AI Act requires employers to tell workers before a high-risk AI system is used on them at work. That covers the systems that hire, assign, monitor and fire. It is the only binding rule of its kind. In Germany, a works council, the body employees elect to speak for them, must be told of plans to introduce AI. In Spain, works councils have a right to know how the algorithms that affect jobs work.

Australia. A first-of-its-kind report from the Department of Employment and Workplace Relations, AI and employment in Australia, found in July 2026 that AI was not yet causing mass layoffs, as ABC News in Australia reported it.

The United States has, so far, one state law that asks about AI on a layoff notice. Our piece on the law sets it out.

Abroad, as at home, companies name AI as the plan, or deny it. Our piece on what the companies say sets out that pattern in America.

Lufthansa said it would shed 4,000 jobs by 2030 "with the help of artificial intelligence."

Barclays is cutting about 5,000 roles, 5 percent of its staff, most of them in support functions. Its own study in September flagged computer and engineering roles as the most exposed to AI. Whether that study drove the cut is not established. Its chief executive told analysts that AI is "the icing on the cake, not the cake."

Japan's Mizuho was widely reported to be replacing 5,000 clerical jobs with AI. Nikkei, which broke the story, described something else: shrinking an administrative staff of about 15,000 by up to a third over ten years, by moving people and hiring fewer.

There are two ways to read this, and the evidence supports both.

The first is that the reversal has started. The Philippines has cut its own forecast. Ireland's tech jobs fell by a tenth. Britain's graduate vacancies are at a low not seen since 2016. In South Korea, almost all the jobs the young lost were in the sectors most exposed to AI.

The second is that most of this is still exposure and forecast, not layoffs. The Philippine industry's own chief executive names AI as one cause among several. Ireland's fall is blurred by a change in how it is counted. India's industry body says its workforce is still growing. Australia's government found no mass layoffs yet.

Either way, two things follow.

First, this is the first technology shock where the rich world's safety nets will be tested by their own middle class. The office workers, the graduates, the women in clerical jobs. Those countries have the money to respond, if they choose to.

Second, the countries that climbed by doing desk work for the rich world may lose that ladder before they finish climbing it. For the Philippines and India, the question is not whether AI arrives but whether the next rung is there when it does.

Our ask is simple. When you hear that AI is coming for jobs, ask whose jobs, and in which country.