Apple Acknowledges U.S. Senate’s Request for Details about iPhone X Face ID Feature
As in previous years, Apple’s 2017 September keynote brought device announcements, new feature introductions, and other headline-worthy events. What’s different this year, though, is that it also prompted the United States Senate to contact Apple about security and privacy concerns.
Those concerns surrounded the announcement of Apple’s revolutionary iPhone X, the first iPhone ever to ditch the commonplace home button. Specifically, the Senate had questions about the new iPhone’s Face ID feature, which will replace Touch ID as the primary way to unlock the iPhone.
The letter, penned by U.S. Senator Al Franken and addressed to Apple CEO Tim Cook, asked questions about how Apple will use facial recognition data. For instance, will the company use that data “to benefit other sectors of its business,” or to sell to third parties? Franken also inquired about Face ID’s ability to recognize a diverse range of faces, and about whether you could hoodwink Face ID by someone holding the iPhone X up to the owner’s face.
One month later, Apple has formally responded to Franken’s letter. The response—signed by Apple’s Vice President for Public Policy, Cynthia Hagan—acknowledges each of Franken’s concerns.
Hagan says that Face ID uses an on-device system called Secure Enclave to process facial data. This data, Apple assures, is fully encrypted and does not leave the device at any point during the processing stage. In other words, Apple doesn’t handle the data at all—let alone store it for later use.
Regarding bias, Hagan said that making Face ID accessible to people of all races and ethnicities was a high priority. The company developed the software using over a billion images. The development and testing process, Apple says, included a “representative group” of different ages and ethnicities.
Finally, regarding Face ID “spoofing,” Hagan assures Franken that it won’t be easy. The software uses gaze detection to read attentive faces, making it more difficult for someone to unlock a phone by holding it up to the owner’s face. Breaking into a phone when someone is unconscious or under duress, in other words, will be harder than it seems. Hagan also noted that the neural network behind the Face ID feature has been carefully trained to tell the difference between real faces and things like photographs or masks.
You can read Hagan’s full response letter here.