|
Briefing
|
- AI-Generated Fingerprints – Researchers from New York University used neural network to generate fake fingerprints called DeepMasterPrints that can be used on biometric scanners
- 23% Replication Rate – Was able to replicate 23% of fingerprints in biometric identification system, with error rate at one in thousand
- AI Training – Trained generative adversarial network AI with dataset of real fingerprints
- Insecure Fingerprint-Based Systems – Current biometric scanners only scan partial fingerprints with portions of fingerprints potentially similar to others
- Vulnerable Multi-Fingerprint Biometrics – Can be used to infiltrate biometric systems with many fingerprints on record, as opposed to phone-based fingerprint scanners where there are only few prints stored, as neural network has more chances of matching partial fingerprints to many existing in database
- More Secure Systems – Researchers hope their work can help make biometric systems more secure
|
|
Accelerator
|
|
|
Sector
|
Information Technology
|
|
Organization
|
New York University (NYU)
|
|
Source
|
|
|
Original Publication Date
|
November 15, 2018
|
