Skip to Content
logologo
AI Incident Database
Open TwitterOpen RSS FeedOpen FacebookOpen LinkedInOpen GitHub
Open Menu
発見する
投稿する
  • ようこそAIIDへ
  • インシデントを発見
  • 空間ビュー
  • テーブル表示
  • リスト表示
  • 組織
  • 分類法
  • インシデントレポートを投稿
  • 投稿ランキング
  • ブログ
  • AIニュースダイジェスト
  • リスクチェックリスト
  • おまかせ表示
  • サインアップ
閉じる
発見する
投稿する
  • ようこそAIIDへ
  • インシデントを発見
  • 空間ビュー
  • テーブル表示
  • リスト表示
  • 組織
  • 分類法
  • インシデントレポートを投稿
  • 投稿ランキング
  • ブログ
  • AIニュースダイジェスト
  • リスクチェックリスト
  • おまかせ表示
  • サインアップ
閉じる

レポート 496

関連インシデント

インシデント 3221 Report
Identical Twins Can Open Apple FaceID Protected Devices

Loading...
The iPhone X's Face ID can be fooled by identical twins
mashable.com · 2017

The iPhone X has a weakness for identical twins.

It can’t really tell the difference.

It's the first Apple device to include Face ID, a face-mapping technology that can be used unlock the phone, verify Apple Pay, and essentially replaces the fingerprint scanner (or Touch ID).

On the face of things, this trade-off makes perfect sense. Apple’s Face ID, according to the company, is more secure than Touch ID. Face ID has a 1-in-1 million false acceptance rate (or identifying someone else as you), as opposed to Touch ID, which has a 1-in-50,000 false acceptance rate.

Apple’s Face ID also proved to be, in my tests, a powerful and consistent hands-free iPhone unlocking strategy. It was very good at recognizing me, even when I wore a hat or a wig.

When Apple unveiled Face ID in September, it did warn, however, that its 1-in-1 million false acceptance rate might be somewhat lower if presented with two people with very similar DNA. In other words, siblings or identical twins gave the system problems.

There are no good numbers for exactly how many identical twins there are in the world, just an oft-trotted out statistic that 32 out of every 1,000 people is a twin. Even as multiple birth numbers rise, the numbers for identical twins are likely lower.

Based on those sketchy stats, maybe it would’ve been unwise for Apple to design Face ID to beat the twin test. Even so, some of us know enough identical twins (I’m looking at you Property Brothers) to wonder if the iPhone X’s Face ID technology would work for them.

Seeing double

At Mashable, we’re blessed with access to two sets of identical twins. Each agreed to bring in his twin, sit with us, and put Face ID to the test.

Both twin sets are brothers: MJ Franklin and his brother Marc, and Carlos Cadorniga and his brother Alex. Each twin set shared how they often confuse friends and family — I still have trouble telling one set apart. Could the iPhone X's Face ID tell the difference?

To test Face ID’s Twin-ID-ing capabilities, we had one brother register his face on the iPhone X, verify that he could unlock the phone by looking at it and then hand the locked device to his identical twin brother.

Marc and MJ Franklin (or is it MJ and Marc?) take the Face ID twin test. Image: Lili sams/mashable

With both sets of twins, the other twin unlocked the iPhone X, even though neither one had registered his face with Face ID on the iPhone X. With the Franklin twins, we had both brothers remove their glasses and had the other brother register. Again, Face ID failed to tell the difference.

Look, Apple never claimed Face ID was perfect and, in my tests, it could not be fooled by photos or videos of my registered face. Still, these results do not bode well for all the identical twins out there, to say nothing of triplets and quintuplets. This is, by the way, a test Microsoft says its Windows Hello Facial recognition technology reportedly didn’t fail.

Since Face ID is backed by powerful silicon and algorithms — it even learns how your face changes over time — we can only hope that Apple will continue to strengthen Face ID’s twin-discerning capabilities. In the meantime, identical twins will probably be using a passcode on the iPhone X.

情報源を読む

リサーチ

  • “AIインシデント”の定義
  • “AIインシデントレスポンス”の定義
  • データベースのロードマップ
  • 関連研究
  • 全データベースのダウンロード

プロジェクトとコミュニティ

  • AIIDについて
  • コンタクトとフォロー
  • アプリと要約
  • エディタのためのガイド

インシデント

  • 全インシデントの一覧
  • フラグの立ったインシデント
  • 登録待ち一覧
  • クラスごとの表示
  • 分類法

2024 - AI Incident Database

  • 利用規約
  • プライバシーポリシー
  • Open twitterOpen githubOpen rssOpen facebookOpen linkedin
  • e1b50cd