Facial recognition has been widely used by authorities to identify alleged 'criminals' in a specific court case. It has been the basis of most arrests in the United States. Now, the technology is being scrutinized for its failed system towards correctly identifying who's the suspect and who's not. And that issue also applies, even to our politicians. 

How bad is facial recognition?

As first reported via Threatpost, over 100 politicians-- both from the United States and the United Kingdom governments-- are mistagged as 'criminals' with the usage of Amazon's facial recognition. 

Paul Bischoff, a consumer privacy expert and the editor of the 'pro-consumer website' Comparitech, made interesting findings proving how bad is the facial recognition used by authorities.

In his May 28 blog, he told everyone to not fully trust the accuracy of Amazon's Rekognition after his team found the faces of 73 U.K. politicians and 32 U.S. Congressmen identified as law-breakers with pending cases in the court. How did it happen?

As explained, the Rekognition was used to compare both images of 1,959 lawmakers in both countries, from the images of 25,000 randomly selected arrest photos from the Jailbase website. 

The experiment was repeated only once each image, and the results were tallied averagely. Comparitech used an 80 percent confidence threshold to process the images. Rekognition, however, incorrectly identified the faces of the arrested criminals from the politicians.

However, when the confidence threshold was made higher to above 95% matches, the results ended up into zero matches. 

"Be skeptical any time a company invested in face recognition peddles metrics about how well it works. The statistics are often opaque and sometimes downright misleading," said Bischoff on the blog. 

Facial recognition is also 'biased'

Aside from proving how inaccurate facial recognition was, another theory that they prove is how racial bias is these types of technology. 

Data shows that out of the 12 politicians who were misidentified at a confidence threshold of 90% or higher, six were politicians in 'color.' This shows that facial recognition has misidentified more non-white people compared to white people. 

"Putting too much faith in face recognition can lead to false arrests," said him. "But whether you agree with police use of face recognition or not, one thing is certain: it isn't ready to be used for identification without human oversight."

Amazon, Microsoft, Google refuse to sell facial recognition to police

Recently, leading tech companies Amazon, Microsoft, and Google decided to refuse to sell facial recognition to police departments after cases of misidentified criminals in their database.

Amazon specifically announced a one-year moratorium from using the technology since the racial arrests were held with the usage of their technology.

ALSO READ: Disney Is Using Deepfakes and Facial Recognition To Bring Back Dead Actors