How-to
How to know when a dataset's license or terms change
A dataset you're already using can quietly change its terms — commercial use revoked, redistribution restricted. Nobody tells you. Here's the pattern for catching it.
3 steps
- Record the current terms text and a hash of it — Not a summary — the actual declared license/terms text, hashed so any real change is detectable, not just a vibe.
- Re-check on a schedule — Re-fetch the same terms URL and re-hash — a changed hash means a real, verifiable change happened, not noise.
- Get a risk classification with evidence, not a legal opinion — Derivative-work rights, PII exposure tier, both computed from the declared text — always with the disclaimer that it's risk classification, not legal advice.