BBC uses facial recognition to identify terrorist as police officer
In the aftermath of a series of terrorist attacks against people in Israel, one media outlet attempted to identify perpetrators with facial recognition. The practice raises several contentious questions about the ethics, propriety, and practicality of journalists or others investigating the identity of individuals with biometrics.
The BBC reports that one of the terrorists caught on video footage was identified with a facial recognition as a Palestinian police officer. The outlet used Amazon Rekognition, and says it got a similarity score of 94 to 97 percent.
The article provides the caveat that “some campaigners, however, have raised concerns that non-white faces can be falsely identified on facial recognition tools.”
Amazon recommends that that input images for facial recognition searches not include headbands or open mouths, both of which are present in the image used for identification. The image also appears contrary to other recommendations, like avoiding shadows.
“Like any security technology, facial recognition has strength and weakness,” Biometrics Institute Chief Executive Isabelle Moeller tells Biometric Update in an email. “It is important to carefully consider the benefits and risks of using this technology before deploying it in any context, and to take steps to mitigate the risks and potential harms.
“The Biometrics Institute would always evaluate this type of unofficial deployment of face recognition technology by exposing the use case to its Good Practice Framework and other assessment tools that it has developed in recent years,” Moeller adds. “In situations where facial images are captured by members of the public it is essential to verify the provenance of the images and establish that they are genuine and have not been altered in any way, such as a deepfake. This is true of any such data that could potentially be used in a court of law, and this is a basic requirement for any law enforcement agency that produces evidence from any source. “
Moeller urges users of facial recognition to “always consider and ask” about the provenance of the facial image database being searched, and how the authenticity of the images can be established and how they can be used in compliance with the appropriate laws and regulations. Organizations should also be aware of how accurate the algorithm used is, and ask “has the generated result been evaluated by a qualified and accredited face recognition practitioner?”
A final important question, Moeller says, is: “What are the specific benefits of using facial recognition technology in this context, and how do they outweigh the risks and potential harms?”
“Unless these issues can be resolved satisfactorily any search through such a system would be inconclusive and fundamentally untrustworthy,” she cautions.
The BBC had not responded to questions about how and why the facial recognition search was performed as of press time.
Article Topics
Amazon | BBC | biometric identification | biometrics | Biometrics Institute | facial recognition
Comments
One Reply to “BBC uses facial recognition to identify terrorist as police officer”
Leave a Reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
That was incredibly unethical on the part of the BBC, a decision that police training in biometrics would not have allowed.
So, presuming the BBC users understood what the biometrics actually mean (a good chance that is not the case here) there is a 3 to 6 percent chance the BBC called an innocent man a terrorist – and they did this publicly, to an audience of tens of millions.
They should have reported this to the authorities – that would have been the smart thing to do – but making a public story out of this, and creating such a huge risk for someone who is potentially not the person they want him to be – that is profoundly unethical. They may have ruined the life of an innocent man, and for what? A story.
This should have been passed to the police for investigation. It should not have been broadcast internationally.