Explains protecting production data in test environments
Masking, pseudonymisation and anonymisation (ISO/IEC 20889, GDPR recital 26); refreshes approved by the owner.
Can AI do this?
AI does this well Most of the tested AI models master this. 3 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 99% | Mastered |
| gemini-3.5-flash-lite | Low cost | 91% | Not shown |
| gpt-6-luna | Low cost | 99% | Mastered |
| claude-haiku-5-5 | Low cost | 98% | Mastered |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: Masking, pseudonymisation and anonymisation (ISO/IEC 20889, GDPR recital 26); refreshes approved by the owner.
How it is assessed: Answers a written question on the test data protection correctly
Related knobits
Builds on
Needed for
- O-201 Recognises production personal data in non-production
- O-202 Recognises re-identifiable masking
- O-208 Recognises an unapproved production refresh
- O-231 Replaces re-identifiable masking with protection fit for non-production
All 15 knobits of Plan environments and capacity.
Source: Practice the stage 2 reference and review checklist rest on