Privacy Lab Interactive
Techniques the framework uses so intelligence can be shared without exposing people. Use fake values only.
1 · Edge tokenisation
Plain SHA-256 (guessable for small ID spaces)
Keyed token (key never leaves the source institution)
A plain hash of a structured ID can be brute-forced, so the design uses keyed tokens created at the edge.
2 · Differential privacy
True count of mule accounts in a region: 1284
Published (Laplace noise): 0
Lower ε adds more noise and more privacy; re-slide to resample.
3 · Selective disclosure
| Name | A. R*** | hidden |
| Account | ••••4471 | masked |
| Risk band | High | shared |
| Fraud vector | SIM swap | shared |
| Evidence fingerprint | sha256:9f2c… | shared |
Partner states receive only what they need, plus a hash to verify authenticity.
4 · Right to erasure by key destruction
Personal data stays encrypted off-chain. Destroying the key makes the data unrecoverable while the on-chain fingerprint remains as an integrity record without personal data.
Conceptual; consult counsel on erasure obligations for blockchain-anchored records.