At a time when there is so much fear about the unethical use of facial recognition, AnyVision is in talks with the Israeli defense technology company Rafael Advanced Defence Systems for a joint venture. This follows Microsoft’s decision in March this year to not invest in AnyVision after reports suggested that the Israeli company’s technology was being used for surveillance in West Bank.
Large companies have, for the most part, stayed away from being part of any venture that promotes the use of facial recognition in mass surveillance. Amazon and IBM had also announced decisions to either restrict the use or not get involved in facial recognition for public purposes.
AnyVision’s decision goes against such an ethical stand that more and more companies wish to promote and drives straight into the ongoing argument among companies, the public, and governments about regulating face recognition technology. This brings three major concerns into the focus.
Will more face recognition startups follow suit?
The biggest fear is that this could be a tipping point where more and more face recognition companies, especially startups desperate for funding, begin to allow their solutions to be used for defense purposes. Following Microsoft’s exit from a deal, AnyVision had reportedly suffered a financial crunch and had fired several employees while cutting salaries of several others.
AnyVision does claim that its operations in the defense sector will remain separate from its commercial business. But regardless of their reason for this, a deal like this is just what a company in distress would need to turn things around. Perhaps as a confirmation on investor confidence rising after this deal, AnyVision even announced a fresh round of funding worth $43 million this month.
Foolproof identification system
Besides such concerns on the direction of face recognition companies, many experts are also not convinced of the accuracy of the technology. A report from the U.S. National Institute of Standards and Technology, based on evaluation of 189 algorithms, showed that most programs offer varying levels of accuracy depending on demographics, including sex, age, and racial background. NIST had also said,
in a separate report, that things like face mask affect the accuracy of facial recognition.
Such inaccuracies could result in wrong identification, and in countries where racial violence is common, it could fuel biased decisions and convictions. For instance, the study showed higher rates of false positives for African American females than for any other group. Given the frequent issues of police violence against black people in the U.S., such an error could prove extremely problematic.
Does it invade privacy?
Finally, facial recognition is a significant privacy invasion for many people. Several governments agree with this point of view and have taken steps to ban the use of facial recognition for public security.
Authorities and independent agencies in many countries have also come up with guidelines on the ethical use of facial recognition. In the U.K., for instance, Biometrics and Forensics Ethics Group (BFEG) has issued guidelines asking facial recognition developers to diversify their data sets. Developers are encouraged to follow these guidelines and tailor their solutions in such a way as not to invade the privacy of the citizens.
Endnote
Facial recognition technology is
improving at a rapid rate, but the market has not evolved fast enough to adopt it. But the current situation where many companies refrain from investing in the technology and COVID-19 related financial concerns plague earnings, face recognition companies may open up to funding sources that they may have avoided before. After AnyVision’s decision to join hands with Rafael, the market would keep a close watch on any other deals that might follow.