Guide for ethical, legal use of facial recognition published by BSIA
A 16-page document detailing parameters for the ethical and legal use of automated facial recognition (AFR) by companies has been published by the video surveillance section of the British Security Industry Association (BSIA).
The body said in a press statement that the report, which it touted as an industry-first, contains useful terms and abbreviations, ethical uses and a specific focus on distinctive biometric application types such as verification and identification.
Composed by the BSIA working group on automated facial recognition, the report was produced subsequent to recommendations by the Organisation for Economic Co-operation and Development (OECD) regarding the responsible use of artificial intelligence.
The report vividly presents what AFR is all about and explains the key factors which companies have to take into consideration before choosing a facial recognition solution for their biometric security systems.
The guide, which is already available for download, also seeks to enable industry experts and the general public understand the AFR framework at all levels.
Dave Wilkinson, director of Technical Services at the BSIA, who led the process to come up with the guide, explained its importance, calling it the result of a piece of “collaborative work among industry experts.”
“The use of AI is an exponentially growing part of daily life and we must ensure that all stakeholders are aware of the ethical and legal considerations of using these solutions,” Wilkinson said. “If not, this beneficial technology could be misused, leading to loss of trust and increased skepticism of the technology.”
He added: “We want to make sure the general public know that this ethical and legal guidance is out there for companies to follow. Compliance with the law is paramount using this technology, and this guide will provide companies with the basis to demonstrate their commitment to complying with the ethical realities, consequences and impacts of using an AI/AFR solution.”
Value-based principles recommended
The guide notes that although there is no single global ethical framework for the safe use of AI, there are however a number of value-based principals which must be taken into consideration when using the technology.
One of the issues highlighted by the guide is that AFR should be deployed in a way as not to cause any harm whatsoever to individuals’ privacy, dignity or human rights.
Per the guide, there is need for necessary risk assessment, and data protection impact assessments (DPIAs) must be carried out before using facial recognition, and companies must also give relevant training to their staff who may need to authenticate AFR verification.
Another key point mentioned in the guide is that facial recognition should not discriminate. It says that any inaccurate decisions that may be perceived as biased in the AFR system should be reported to the biometric technology provider which should, in turn, take the appropriate corrective action.
In the same vein, the document suggests that the choice of software to analyze the captured image when using a facial recognition system should be carefully considered if such biases must be checked. The guide goes on to recommend that companies should in this wise consult internationally respected organizations such as NIST in the United States, which publishes results of its tests online.
Another recommendation is that facial recognition should be used transparently with the possibility of an individual consenting to the use of their personal data. Training data, it adds, should also be obtained lawfully and within overarching ethical and legal prescriptions.