Meta's Facial Recognition Empire: From DeepFace to a $1.4 Billion Privacy Settlement
How Meta built the world's largest facial recognition database and the landmark Illinois lawsuit that forced its dismantlement
For nearly a decade, Facebook operated one of the most powerful facial recognition systems on the planet. The company's DeepFace algorithm, developed in 2014, could identify human faces in photographs with 97.35% accuracy — approaching human-level performance. By default, the system automatically scanned every photo uploaded to Facebook, identified faces, matched them against its database, and suggested tags. The result was a facial recognition database containing the biometric faceprints of billions of people, built without the explicit, informed consent of those being cataloged.
The legal reckoning for this practice came from an unexpected source: Illinois' Biometric Information Privacy Act (BIPA), passed in 2008 before Facebook's facial recognition system existed. BIPA requires companies to obtain informed consent before collecting biometric identifiers, including facial geometry.
Key Takeaways
- Meta's DeepFace algorithm achieved 97.35% facial recognition accuracy and built a database of over one billion faceprints without informed consent
- The Illinois BIPA lawsuit resulted in a $650 million settlement with eligible users receiving approximately $400 each
- Meta shut down its facial recognition system in 2021 but did not commit to never reintroducing the technology