Integrity that needs an enforcement deadline isn't integrity

Someone built the risk model that treated dual nationality as a fraud signal. Someone joined the datasets, someone kept the scores running, and the damage ran for years before a court or a data protection authority got near it. Nobody in that chain needed an AI Act to know better.

I want to make an argument I know is unpopular in my own field: the data profession has spent a decade waiting to be regulated, and regulation was the wrong thing to wait for.

The damage didn’t look like AI

Three cases, all Dutch, all documented.

In the childcare benefits scandal, the Belastingdienst processed dual nationality as a fraud indicator. The Autoriteit Persoonsgegevens ruled that processing unlawful and discriminatory in 2020. A parliamentary inquiry later tied it to tens of thousands of wrongly accused families, and a cabinet resigned over it. SyRI, the welfare-fraud profiling system, was struck down by the District Court of The Hague in February 2020 for breaching the right to private life under the European Convention on Human Rights. In Rotterdam, a welfare risk algorithm scored people higher for things like poor Dutch and single parenthood — surfaced in 2023 by journalists at Lighthouse Reports and WIRED, not by anyone inside the building.

In every one of these, data professionals were present at every step. And every time, the stopping power came from outside: a regulator, a court, a reporter. Never the room where the model was built.

Self-regulation isn’t a fantasy

It can go the other way. Amazon built a recruiting model that taught itself to downgrade CVs containing the word “women’s,” and the company caught it and scrapped it internally before it ever shaped a public hiring decision (Reuters, 2018). That is what self-regulation actually looks like: someone holding a standard, and checking the output against it.

So the question almost asks itself. Why does a profession this consequential still have no standard to check against?

Eight years of drafting

The idea of a Hippocratic oath for data work has been circulating since a 2018 workshop that produced twenty draft principles and a hope for a hundred thousand signatures. An academic paper at AMCIS 2025 — “An Oath for the Data Professional” — was still proposing what the core values should be: dignity, interpersonal trust, justice, self-control, autonomy, reliability, responsibility, trustworthiness. Eight years, and the field is still drafting.

Compare that to professions that decided not to wait. Medicine had its code before any government regulated healthcare. The Bar governs itself. Accountancy sets standards that reach well past tax law. When Brussels — no one’s idea of a fast-moving institution — regulates faster than your profession can agree on its own conscience, the problem stops being Brussels.

Where regulation stops short

Here is the part I want to be careful about, because it is where the argument usually gets misread. The AI Act is real, it is necessary, and its high-risk categories — recruitment AI, credit scoring, biometrics — deserve every line of it. The deadline this August is not the enemy.

All true. But waiting for regulation and loudly calling for regulation do the same quiet thing: they park accountability somewhere other than the person joining the datasets. And regulation, by design, only reaches the edge. The majority of daily data work — which number sits at the top of the dashboard, how a three-month decline gets framed, what context gets added and what gets left out of the board report — sits outside any framework, now and probably for good. Those are judgment calls. No regulation will answer them, and most wouldn’t try.

Integrity that needs an enforcement deadline is not integrity.

What this looks like on a Tuesday

I am not asking anyone to sign a manifesto. I am asking for a smaller and harder thing: that the benchmark for taking our work seriously stops being “are we compliant?” and starts being “are we holding ourselves to a standard we would be comfortable naming out loud?”

There is a paper proposing eight values for our profession. Read it, then ask yourself the only question that matters: could you sign it today, and mean it?

Happy to think this through with you.