Artificial intelligence is no longer a distant prospect: in 2026, it is moving into offices, factories, hospitals and public services. For millions of French workers, the question is no longer theoretical. It is concrete, immediate and sometimes anxiety-provoking. What is really happening? Which occupations are genuinely in the firing line, and how should people respond?

5 million jobs at risk: what does this figure really mean?

The figure has been circulating for several months: according to a study by Coface and the Observatory of Threatened and Emerging Jobs, AI could affect up to 5 million jobs in France. It is an impressive number, but one that must be read carefully.

The researchers clarify that the figure “reflects exposure of tasks, not their destruction.” In other words, it does not mean that 5 million jobs will vanish overnight, but that 5 million positions include some tasks that can be automated. According to available analyses, around 16% of professional tasks are currently exposed to automation.

Of the 923 occupations analysed, 120, or 13%, show a high level of exposure, with more than 30% of their tasks potentially replaceable by AI tools. The entire labour market is therefore not in immediate danger, but a significant share of certain occupations is.

The occupations most exposed in 2026

Contrary to what might have been imagined ten years ago—when automation was expected to hit factories and warehouses first—it is knowledge work with a strong repetitive component that is now on the front line.

The occupations most exposed in 2026 notably include:

• Data-entry operators and call-centre agents: call centres have seen job vacancies fall by 67% in two years.
• Copywriters: vacancies dropped by 53% over the same period under the influence of generative AI.
• Generalist translators: they face strong competition from tools such as DeepL and multilingual models.
• Junior accountants and administrative assistants: data entry, filing and reporting tasks are widely automatable.
• Bank customer advisers and contract-analysis lawyers: analysis and document-processing tasks are exposed to automation.
• Junior IT project managers: job vacancies have fallen by 48% in two years.

What stands out is that generative AI is hitting qualified occupations head-on. Engineering shows roughly 29% exposure, legal, financial and creative professions around 27%, and managerial and administrative functions about 24%. The assembly-line worker is not the first priority target: the middle manager is.

But AI also creates jobs—a great many

It would be incomplete—and inaccurate—to see AI only as a destructive force. In 2026, France has more than 166,000 AI-related job vacancies, putting it ahead of other European countries in this field. Vacancies jumped by 156% in one year.

New AI occupations cover a broad spectrum:

• Technical roles: Data Scientist, ML Engineer, AI Developer.
• Creative roles: Prompt Engineer, AI Content Designer.
• Strategic roles: AI Product Manager, AI Ethicist.
• Operational roles: AI Trainer, Automation Specialist.

According to the World Economic Forum's Future of Jobs Report 2025, AI is expected to create 170 million jobs by 2030, for an estimated net gain of 78 million additional positions worldwide. Destruction is not the only possible outcome.

How can you adapt in practical terms?

Faced with this dual movement—the disappearance of certain tasks and the creation of new occupations—career change is becoming central. The good news is that moving into AI-related work does not necessarily require five years of computer-science education.

According to several training organisations, 3 to 6 months are enough to qualify for AI-augmented roles when starting from a well-established professional background. Jobs such as Prompt Engineer, AI Consultant or AI Trainer are accessible to nontechnical candidates through targeted training.

Several funding schemes can support this transition:

• The CPF personal training account: it can be used directly for certified AI courses.
• Career transitions financed by OPCO skills operators: for employees whose occupation is under threat.
• Transition Pro: retraining leave lasting up to 24 months, with part of the salary maintained.

The real question: which skills should become indispensable?

Beyond formal retraining, the best protection is to incorporate skills that AI still struggles to reproduce: situational judgement, authentic human relationships, applied creativity and the ability to frame a problem—the point at which AI excels at solving problems that have already been formulated.

A doctor who knows how to use an AI diagnostic-support tool is more effective than one who ignores it—and difficult to replace. A lawyer capable of challenging AI-generated summaries is more valuable than one who merely produces the summaries personally.

The transformation is under way. It is fast and sometimes brutal for certain sectors. But in 2026, a window remains open to adapt, train and reposition your skills in a labour market undergoing profound change.