AUP false positives on government accountability / open data analysis

Problem

When using Claude Code to work on a Dutch government spending transparency project (dutch-gov-accountability), I repeatedly hit AUP (Acceptable Use Policy) violations for completely legitimate queries.

The project collects publicly available open data from:

• CBS (Centraal Bureau voor de Statistiek) — Dutch national statistics
• Rijksfinancien — Central government budget data
• Iv3 (dataderden.cbs.nl) — Municipal finance data for all 342 Dutch municipalities

All data sources are official government APIs designed for public transparency and accountability.

What triggers it

Even simple investigative queries like "where is the culture money going" (i.e., querying municipal spending on taakveld 5.4 Musea) trigger AUP violations. The intent is analyzing public expenditure data that the Dutch government publishes specifically for civic oversight.

Expected behavior

Analyzing publicly available government spending data for accountability purposes should not trigger AUP flags. This is exactly the kind of civic transparency work these datasets are published for.

Workaround

Phrasing queries in more technical/database terms ("query iv3_municipal_finance for taakveld 5.4") instead of natural language ("where is the culture money going") seems to reduce false positives, but this shouldn't be necessary.