Navigating the Biometric Anatomy
summary
A Microsoft legal team builds a review framework: a working definition of biometric data, an inventory of uses, and risk buckets keyed to unique identification versus mere face detection, plus pivots for one-to-one vs one-to-many matching and geography. Requirements include written consent for identification, retention/destruction rules, vendor disclosure controls, and edge-first processing. Litigation trends: exploding BIPA suits, challenges over photos, voiceprints and vendor liability, and regulators (FTC, Canadian DPAs) ordering deletion of raw data and derived models. Federal US law is unlikely soon; definitions may converge toward GDPR/Virginia-style tests.
AI-written summary · about
chapters (6)
- 0:00Start
- 19:50Discover your uses - Distribute your definition
- 24:51Determine requirements based on uses - Can be a few pivots - Unique identification or not - 1:1 matching y. In matching
- 27:04Design your requirements along pivots
- 29:25Think about peculiar risks that might arise -Certain peos do not distinguish between processors and controllers
- 34:25Scope of biometric ID and information - Unique identification
related talks
37mBiometrics & Privacy: Time to Faceoff or is that FaceApp?
43mMade to Measure? The Biases and Boundaries of Biometrics
25mCloud Enabled Biometric Data Collection
19mGDPR, AI & use of Video data
01h 15mHOW TO RECONCILE FACIAL RECOGNITION TECHNOLOGIES WITH CONSUMERS' PRIVACY
47mFace/Off: The Battle to Authenticate with Biometrics
39mHot Legal Issues
01h 11mLiquid AI: Accountability and governance of transnational AI systems.