Facial recognition algorithms can struggle to recognise individuals as the same person after as little as five years.
This means that systems that rely on facial recognition may need to obtain new photographs of users on a regular basis or risk being unable to identify them, according to News Scientist.
While facial recognition systems are capable of up to 99.97%1 accuracy and offer tremendous potential, it is still critical to be aware of their drawbacks in order to protect personal information.
So say cyber security experts at Ping Security. “There are several advantages of facial recognition technology for society. The DeepFace software from Facebook, which has a 97.252 accuracy rate for recognising human faces, is used for security and crime prevention purposes.”
In addition, they say facial recognition technology can also improve efficiency during tasks such as border checks at airports by reducing the need for interpersonal contact, and in the field of medicine, it can be used to detect genetic illnesses by detecting subtle facial features.
However, it is not without its risks. “As facial recognition technology advances, cyber thieves try to exploit it. In 2019, bad actors cracked Apple iPhone’s FaceID user authentication in approximately two minutes,” they added.
Therefore, although the advantages are many, the dangers and drawbacks need to be weighed too, in order to keep information secure.
Firstly, the researchers say it can violate individual and societal privacy. Because facial recognition algorithms may store significant amounts of data, it is critical that they employ the highest levels of security. It is also key for vendors who engage with these companies to have adequate security procedures in place to eliminate any backdoor cyber security risks.
It also creates data vulnerabilities. “Facial recognition information could fall into the wrong hands as a result of data breaches. Although systems are improving in their ability to prevent identity theft, it is still possible.”
In addition, organisations may use the information provided to them for research purposes, and as a result, stand to profit from it without permission.
Companies might also share or even sell this data to third parties, enabling them to recognise and follow individuals. “Companies developing face recognition apps should consider data privacy risks. One method is to create detailed privacy policies, obtain customer consent, and provide them with the option to opt-out of the programme,” the company adds.
Another drawback is that to make face matches, the technology relies on algorithms. Because the databases contain more data on white males, the algorithms are more robust for white men than for women and people of colour. “As a result, the algorithms develop inadvertent biases.”
Zain Malik, senior product marketing manager from Ping Identity, is hopeful though that as face recognition technology are constantly being improved, "the percentage of errors
Companies using facial recognition systems
- Facebook’s DeepFace technology has a true positive rate of 97.25%.
- Google’s FaceNet was 99.63% accurate when matching 13,000 pictures of faces from across the web.
- Amazon’s Rekognition is a cloud-based facial recognition service.
- Microsoft’s Face API is another cloud-based facial recognition service.
- Gemalto’s Cogent Live Face Identification System recognises faces in busy environments, allowing developers to create applications that match live faces with data from documents.
Share