Allianz Partners, the assistance and travel insurance subsidiary of the German giant, is set to cut between 1,500 and 1,800 jobs in Europe, out of a workforce exceeding 22,000 people. The stated reason is explicit: the rollout of artificial intelligence. The approach, however, contrasts sharply with what we see across the Atlantic: six months of negotiation with employee representatives, voluntary redundancies and early retirement across five European zones.
We recently discussed Meta's 8,000 job cuts, justified by the efficiency gains of AI. Here is the European version of the same phenomenon, and the comparison is instructive. For while the upheaval is identical, the way of navigating it is not at all.
What was announced
At an event in Munich in early July 2026, Allianz Partners CEO Tomas Kunzmann confirmed the elimination of 1,500 to 1,800 positions across Europe, directly linked to the deployment of artificial intelligence solutions. The news had first surfaced in late 2025, but it is this official confirmation that seals the move.
The roles affected are mainly in assistance and travel insurance activities, particularly in the call centres handling claims and complaints. Voluntary redundancy schemes have been opened in Spain, France, Germany, Italy and the Benelux countries, following six months of talks with works councils. The company says it is supporting the transition through retraining, training and internal mobility.
Why insurance is on the front line
There is a structural reason why this sector is particularly exposed, and it is worth understanding. Insurance rests on exactly what AI feeds on best: highly structured data, essentially textual, governed by precise and repetitive rules.
Processing a claims declaration involves reading documents, checking their compliance with a contract, applying a scale, and producing a reasoned decision. It is work that demands rigour and consistency, more than creative judgement. And that is precisely the type of task where current models excel. The same holds for banking, compliance or underwriting, which explains why other players in the sector, such as Munich Re's insurance arm, have announced comparable reductions in Germany.
Compare the two approaches. On one side, a plan announced overnight, with immediate departures. On the other, six months of negotiation with employee representatives, voluntary departures, early retirement, and support for retraining. The number of jobs cut is comparable in proportion, but the experience of those affected is worlds apart. This is not just a matter of labour law: it is a difference in how a society chooses to absorb a technological shock.
What it doesn't say, and what needs naming
Let's stay clear-eyed about the limits of this comparison. A negotiated transition is still an imposed transition for those who live through it. A voluntary departure is only voluntary insofar as the alternative remains viable, and it is easier to leave at 58 with early retirement than at 42 with a skill that has become less in demand. The social framework softens the blow; it does not remove it.
We should also avoid the analytical error we flagged regarding Meta: AI is not always the sole cause. It changes the productivity equation, making possible reorganisations that other factors, such as margin pressure, had already made desirable for management. Attributing 100% of these cuts to technology is both to credit it with too much power and to erase the human decisions that arbitrated them.
The signal for what comes next
What makes this announcement important is less its scale than what it demonstrates. For years, corporate discourse on AI spoke of assistants, copilots, decision support, rarely of openly acknowledged job cuts. That line has now been crossed, and publicly.
The administrative services sector in developed economies is likely to see similar moves, and projections suggest a significant share of jobs will be affected to varying degrees. The key word being affected rather than eliminated: many roles will transform without disappearing, shifting toward what the machine does not do well, namely the exceptional case, the difficult interaction, the unusual judgement call.
One question remains open, and it is less technological than political: how quickly can a society absorb a transformation of this magnitude, and who bears the cost? Europe is betting on a negotiated pace, slower and more costly in the short term for companies. The United States is betting on speed of adjustment. Both models will be put to the test in the years ahead, and it will be interesting to compare, not just their economic efficiency, but what they leave to the people who live through them.