Senator Markey demands answers from Clearview AI on facial recognition technology

Letter text (PDF)

Boston (November 20, 2023) – Senator Edward J. Markey (D-Mass.), a member of the Commerce, Science, and Transportation Committee, today sent a letter to Clearview AI demanding answers from the company about its facial recognition technology. In his letter, Senator Markey expressed serious concerns about Clearview AI’s mass collection of biometric information without individual consent, arguing that Clearview AI’s practices pose a serious threat to privacy and civil liberties, particularly for people of color. A recent report indicated that a Louisiana sheriff’s office relied on incorrect Clearview AI resulting in the wrongful arrest of a Black Georgia resident who never visited Louisiana.

In the letter to Clearview AI CEO Hoan Ton-That, Senator Markey wrote: “Facial recognition technologies – and Clearview AI’s system in particular – pose a serious threat to privacy rights and civil liberties. Especially in the context of law enforcement use of your product, the American people should not have to give up their privacy for public safety. If Clearview AI is serious about participating in Congress’ work to regulate artificial intelligence, I urge your company to consider the privacy risks of facial recognition technology and be transparent about your company’s development and use of its facial recognition system.”

The senator requests Clearview AI to respond to the following questions by December 11, 2023:

1. Please describe Clearview AI’s process for storing the photos it scrapes from the web and the face prints (or face vectors) it detects on those photos.
2. Please identify the data that Clearview AI used to train its facial recognition algorithms.
3. Does Clearview AI track the race, gender, ethnicity, age and other demographic indicators of individuals in images submitted to and generated by customer searches?
4. Although Clearview AI has submitted algorithms for testing by the National Institute of Standards and Technology, this testing is not designed to replicate real-world conditions. Has Clearview AI tested the accuracy and reliability of its system under conditions similar to real-world use by law enforcement, including use on low-quality probe images and use by poorly trained or untrained law enforcement personnel? If so, was this test performed by an independent auditor? If yes, please provide the results of this test.
5. Please list all of Clearview AI’s public and private customers that have access to Clearview AI’s database, including federal government entities such as intelligence and law enforcement agencies and state and local government entities such as public schools.
6. Please identify all Clearview AI public and private customers who have purchased licenses to use Clearview AI’s facial recognition algorithm.
7. Has Clearview AI ever revoked access to its facial recognition algorithm from a law enforcement agency?
8. Please describe the process a person must go through to remove their facial prints (or face vectors) and photos from Clearview AI’s database.
9. Has Clearview AI changed its cybersecurity practices after the 2020 hack? If so, please describe the changes. If not, why not?
10. Has Clearview AI developed any product that would allow real-time identification of people using their facial recognition technologies?
11. Does Clearview AI attempt to obtain consent or identify another legal basis to collect, maintain or use personal information where required by law?