SIA report debunks facial recognition myths
16 August, 2019
category: Biometrics, Corporate, Government
Paranoia is often the price popularity, and this is certainly the case when you consider the wide range of facial recognition myths. As is often the case, a new technology starts to gain mainstream acceptance, sparking worries, fears and anxiety — and misinformation — about its use. The latest authentication technology battling public misconception is facial recognition. As that biometric authentication method becomes more popular and finds more real and potential uses in the world of digital ID, politicians, regulators and consumers are pushing back, fueled by fears of the technology’s potential misuses.
The analysis from the SIA takes aim at accusations that facial recognition authentication has an inherent racial bias, citing recent research that suggests otherwise
But as the Security Industry Association (SIA) recently pointed out, those fears are often themselves fueled by misconceptions and faulty assumptions. Misconceptions become dangerous when things happen like recent events in San Francisco, where local government has taken steps to severely limit the use of facial recognition in that high-tech city, potentially providing an example for other locations.
Moves to ban or limit facial recognition tend to stem from, among other things, dystopian views of how that biometric authentication method will be misused by government authorities, including law enforcement. As SIA tells it in a new analysis of the myths and facts surrounding facial recognition technology, a common perception is that, “use of facial recognition technology in the U.S. is “out of control” with no safeguards.”
Not so, says the industry association.
Facial recognition safeguards
“Far from a rules-free environment, use of this technology is subject to an existing framework of laws, regulations, administrative rules and best practices that address many privacy and civil liberties concerns,” the SIA analysis states. Among other things, it says, “government use is bounded constitutionally by the First, Fourth, Fifth and Fourteenth Amendments, which prohibit use to suppress free speech and religious expression and protect citizen’s rights to due process and equal protection and against unreasonable searches and seizures.”
As well, the group said, responding to one of the big trends in the digital ID world, “federal law clearly allows states to share driver’s license data, including digital photos with federal agencies, but only for law enforcement and other narrow purposes.”
The growing backlash against facial recognition — arguably part of an even bigger global trend of increasing skepticism about the security of digitally-stored data and personal identifying information — also involves a common view that, according to SIA, “you can be misidentified by law enforcement due solely to facial recognition errors.” But the industry group takes issue with that idea, instead pointing out that “the bottom line is that in investigative applications, facial recognition technology itself does not make a final match determination and therefore cannot identify a person as someone they are not.” SIA adds that a “false positive is not misidentification; it is part of how the process works to create a gallery of potential matches based on a similarity score.”
Debunking common facial recognition myths
The analysis from SIA also takes aim at accusations that facial recognition authentication has an inherent racial bias, stating that facts don’t support that theory. “Recent research suggests that newer algorithms, including many of the top-performing ones tested by NIST, have accuracy rates for African Americans equal to or even higher than those for other groups,” SIA says.
The analysis even pokes holes in the idea that the ongoing backlash to facial recognition technology enjoys popular support. As SIA put it, “in a recent national survey of over 3,000 Americans, only 26 percent believed the federal government should strictly limit the use of facial recognition technology, dropping to 18 percent if limits would come at the expense of public safety.”
The SIA report goes into even deeper detail about facial recognition myths and facts in its online analysis. No doubt these issues will continue to be the subject of intense focus and discussion as facial recognition becomes a bigger part of digital ID and daily life.