Facial recognition offers brands better ways to personalise marketing, but the privacy questions and risks are substantial, warn experts
Facial recognition is carried out by matching faces, whether it might be for emotion tracking in advertising or face matching for criminals, and relies on a dataset of facial images. And this is one of the core issues.
Microsoft recently removed a database called MSCeleb containing more than 10 million facial images, scraped from publicly available sources, after a Financial Times investigation argued the images were taken without consent and included private individuals. In 2018, China and the US accounted for almost all references to this database in research papers and facial recognition projects. A Microsoft spokesperson reportedly said the database was only intended for academic purposes.
In addition, several experiments have found the systems can produce results that are racially bias or inaccurate.
OpenMic has a mission is for greater corporate accountability in the tech and media sector to do with privacy, civil rights and freedom of expression. The group makes the business case for improved or more responsible practices through organising large, institutional shareholders such as funds, companies and investment firms and lobbying to improve protections for civil liberties and highlighting how new technologies can raise the risk profile for companies.
Reputational damage, according to its executive director, Mike Connor, is one of the risk factors for businesses that deploy facial recognition because they jeopardise trust in the eyes of customers and the public in general. The organisation has been active around facial recognition generally and Amazon’s system in particular.
Connor pointed to the lack of permissions to do with the technology and says “one of the most troubling aspects of facial recognition technology is that there are no rules”.
“There are no corporate guidelines, or regulatory or legal rules about how it should be used in the vast majority of situations. Research shows it can be bias and inaccurate. There are many troubling questions,” he says.
OpenMic has been at the forefront of pressure on Amazon to limit how its service Amazon Rekognition is used because of concerns about human rights violations. In defending Rekognition against what it says are misleading claims, the online retail giant argues it has not infringed civil liberties nor is it not bias. Amazon has, however, argued for legislation to protect individual rights and ensure transparency in government use of facial recognition systems.
Still in the US, some cities and states are banning, or considering banning, facial recognition use by government agencies. Even Microsoft's president suggested the US Congress needed to regulate its use and not leave it to companies alone.
As the ethical and privacy debates continue, technology providers are attempting to overcome some of these challenges. One way to improve facial recognition is with better facial data and this can only be done through a dataset used to train the systems accordingly. In this vein, IBM has released a new dataset comprised of more than 1 million diverse facial images in a bid to improve fairness and accuracy in the systems.
The Diversity in Face is a publicly available dataset with 10 different facial coding methods including craniofacial (head length, nose length, forehead height), facial ratios (symmetry), visual attributes (age, gender) pose and resolution. The computing giant argues the breadth of data sets is the key to improving AI-powered facial recognition systems by being better trained.
“For the facial recognition systems to perform as desired – and the outcomes to become increasingly accurate – training data must be diverse and offer a breadth of coverage," researchers said in a blog post.
What's clear is legal protections are needed to safeguard personal data. At the same time, it’s necessary to provide certainty for businesses interested in innovating around facial recognition.
“The legislation is changing all the time around us. And if businesses are making investments in these systems, it’s a matter of working out what they can do today and what they may be able to do in, say, five years time," Ingelbrecht says. "Obviously, if you're making investments in these systems, you want to be sure you can use them.
"It does mean we’re dealing with a moving entity, in terms of regulation, and what we’re seeing in Europe and North America, a progressive catching up of the regulation with the technology.”
Data, while it might hold the keys to better customer intelligence and improved service and better business, nonetheless presents a risk to businesses as well. OpenMic’s Connor echoes concerns and his advice is simple.
“It’s becoming increasingly clear having data is not necessarily a good thing," he argues. "There’s been a rush to ‘big data’ and also AI or they’ll be left behind and, to a certain extent, that’s true. But every piece of data gathered is a potential liability because it raises questions of data security and privacy.
"Good corporate governance is you only want as much data as you need. You shouldn’t be gathering data recklessly because the more data you have, the more you’re exposing your company to potential liability.”
What's more, digital platforms operate in different legal jurisdictions with varying legal approaches such as privacy laws in Australia or GDPR data protection rules in the EU. Proper data protection, such as what’s required in healthcare, might seem onerous to begin with, but Felk says it's also an opportunity to improve information management systems. Still, he concedes it’s a huge challenge to piece together the privacy rules for each country for huge global platforms.
While certain activities such as the capture or sharing or particular bits of information can be regulated, the potential for assembling a rich picture of personal data, linked to facial imagery, poses huge challenges to privacy protection.
“Once you identify a person, connect that with data and start connecting that across other places to form a profile that the person is unaware of, didn’t consent to, doesn’t have the ability to look at or can’t make go away, you've got a concerning issue,” Felk adds.
Follow CMO on Twitter: @CMOAustralia, take part in the CMO conversation on LinkedIn: CMO ANZ, follow our regular updates via CMO Australia's Linkedin company page, or join us on Facebook: https://www.facebook.com/CMOAustralia.
In the third and final episode of our 3-part CMO50 video series exploring modern marketing and why it’s become a matter of trust, we’re delighted to be joined by Telstra’s former CMO and now digital services and sales executive, Jeremy Nicholas, and Adobe VP Marketing Asia-Pacific and Japan, Duncan Egan.
Flash back to the classic film, Willy Wonka and the Chocolate Factory. Television-obsessed Mike insists on becoming the first person to be ‘sent by Wonkavision’, dematerialising on one end, pixel by pixel, and materialising in another space. His cinematic dreams are realised thanks to rash decisions as he is shrunken down to fit the digital universe, followed by a trip to the taffy puller to return to normal size.
Why is it there is no shortage of leadership development materials, yet outstanding leadership is so rare? Despite having access to so many leadership principles, tools, systems and processes, why is it so hard to develop and improve as a leader?
As a nation united by sport, brands are beginning to learn money alone won’t talk without aligned values and action. If recent events with major leagues and their players have shown us anything, it’s the next generation of athletes are standing by what they believe in – and they won’t let their values be superseded by money.
