Amnesty International researchers urge Amazon to stop selling facial recognition to law enforcement

Researchers from industry and academia working on AI have signed an open letter asking Amazon to stop selling its facial recognition technology to law enforcement after discovery of apparent bias .

Palm face: In January, Deborah Raji and Joy Buolamwini of MIT revealed research suggesting that Amazon's Rekognition product misidentified women and people with darker skin more often than other subjects. Buolamwini has already pointed to racial bias in other facial recognition systems (Microsoft has addressed the technology issues identified in its work).

Change of foot: Face recognition has become a major concern for those worried about irresponsible use of AI, and it seems increasingly likely that regulations will be in place. But technology is spreading rapidly and companies are struggling to adjust their positions. Microsoft announced that it would continue to work with law enforcement, but also supported legislation requiring signs indicating where facial recognition is used. Google has stated that it will not provide facial recognition before it can propose an appropriate strategy.

War of words: The letter also refutes critics of a study published in January by two leaders of Amazon, Matthew Wood and Michael Punke. Amazon's rebuttals claimed that the initial search gave an inaccurate picture of the capabilities (and limitations) of recognition. They also noted that Amazon required that its technology be used in accordance with the law and stated that Amazon would support greater transparency of law enforcement by law enforcement agencies. .

Pioneering vision: Among the signatories of the letter are prominent AI and ethics figures, as well as Yoshua Bengio, computer scientist who recently received the million dollar Turing Award with two colleagues for his role in the development of in-depth learning technology that underlies modern AI, crucial element to face recognition. Bengio has recently emerged as a key voice on the risks of AI.