ADVANCE AI Why facial recognition technology in banking is here to stay

ADVANCE.AI: Why facial recognition technology in banking is here to stay?

Return to previous page

The proliferation of technology companies added a new dimension to the usage of biometric technology – especially fingerprint and facial recognition technology. Mobile phones makers took this one step ahead and heavily deployed these technologies across their hardware. Taking note of drastic changes, institutions and businesses especially banks began simplifying their processes, undertook large-scale deployment of technology and strategized for better customer engagement. At the Digital Banker, we believe facial verification will progressively extended to cash withdrawals, deposits, fund transfers, top-ups and bill payments and hence to understand how banks can leverage off this, we reached out to Aradhna Sharma, Regional Director at ADVANCE.AI. 

The Digital Banker (TDB): How can facial recognition be deployed across operations in banking, financial services and even retail and ecommerce?

Aradhna Sharma, Regional Director at ADVANCE.AI

Aradhna Sharma: At banks and financial services as well as in retail and e-commerce, facial recognition enabled by AI and big data are playing major roles across the organisation. Use cases include digital identity verification and eKYC authentication to accelerate remote customer and merchant onboarding. This is especially relevant during Covid19 lockdowns, and to overcome barriers such as geographical distance.

Digibanks in Singapore and across Asia are already using facial recognition as a security measure that reduces the risk of stolen PINs, passwords and ATM cards.

Facial recognition itself is a subset of AI technology, which can be used across a range of scenarios to limit credit and fraud risks, assess alternative credit scoring for underwriting risk or general business transformation and digitisation of manual processes.

TDB: What capabilities can be quickly enhanced by the use of facial recognition technology in the FSI landscape? In the future, how is the banking & financial service segments set to fully benefit from the deployment of this technology?

Aradhna Sharma: Our technology has extensively helped several business initiate customer interaction in a seamless manner. For example, In Indonesia, a top digital lending platform, Danamart, used our facial recognition and liveness detection solution to accelerate their customer verification process with an accuracy rate of over 99 per cent and it was available 24/7.

Danamart was able to accelerate and digitize the entire customer onboarding process, redeploy staff to other critical areas of the business, saving precious time and resources without compromising compliance requirements.

In Jan 2020, the Reserve Bank of India passed regulation allowing banks and other lending institutions in India to use video-based Customer Identification Process (V-CIP) to help them onboard customers remotely. This relies heavily on facial recognition techniques to authenticate, onboard and document new customers. (Here’s an example.)

In Singapore, facial verification features are being progressively extended to cash withdrawals, deposits, fund transfers, top-ups, and bill payments, and in the future, high value transactions or transfers. Hence, the benefit to the banking and financial sector will continue to be significant in the areas of risk mitigation, operational and process efficiency, cost savings and customer experience.

TDB: What are the precautions to ensure the accuracy of such facial recognition technology?

Aradhna Sharma: AI facial recognition technology has to be extensively trained and tested by relevant data to be accurate. For example, Southeast Asian faces, facial and bone structures, eyes and skin tones are very different from say, Caucasian faces, and this is why localisation and extensive testing is required for high accuracy of facial recognition technology in this region.

The technology must also cater for deployment in a variety of low-light or low-resolution situations to accommodate the surrounding environment or lower-end smartphones and camera phone technology.

To prevent misidentification and understand its limitations, new technology must go through rigorous tests and achieve high accuracy before it can be deployed widely. Human intervention also needs to be involved in low probability circumstances or at the right stages to prevent misidentification.

TDB: A subset of AI – how can Facial recognition improve both the security and customer experience within banks/financial services?

Aradhna Sharma: Facial verification and liveness detection can help to quickly verify and authenticate a person’s identity remotely.

It detects high-risk impersonations while also reducing the need for physical visits to the bank, and facilitates processes like video KYC for bank account openings, loan and deposit applications, virtual payments, and remittances. This reduces the need for manual verification while also providing a superior and faster 24/7 solution with better accuracy than a human eye. Manual resources can then be redeployed to other critical areas of the organisation that require human intervention and decisioning.

TDB: Can Covid-19 and the safety guidelines and regulations around it, be a potential hindrance to the progress of facial recognition algorithms and technology?

Aradhna Sharma: Covid-19 and lockdown-related measures have invariably accelerated the adoption of facial recognition algorithms and technology as more people have been forced to turn to mobile or online banking to perform new account openings, loan applications, savings etc. This is both on the consumer end but also internally, organisations have been forced to digitalise many of their previous processes to be able to resume business operations to stay competitive and viable.

With fraud and phishing attacks on the rise in financial services, facial recognition can be a  more secure alternative to bank cards and PIN codes. There is an urgent need for technology that tackles fraud, as the number of online scams and fraudulent transactions in Singapore hit a record high in 2020. The most scams were noted in e-commerce, social media impersonation, phishing and fake loans.

Singapore’s Transport Minister Ong Ye Kung, also a board member with the Monetary Authority of Singapore, told Parliament recently that banks “must do their part” to protect consumers from such losses. This is exactly where we think facial recognition in banking has an important role to play.

TDB: Could you please elaborate on how concerns such as ethics, privacy and security are continuously addressed around the widespread use of facial recognition?

Aradhna Sharma: There are at least three key areas that companies using AI and facial recognition need to keep in mind in terms of ethics. First and foremost, transparency over how data is used, stored and secured. Customer consent is obviously a must. (e.g, through face-to-face consultation that is recorded and documented, or by e-signatures and approvals).

Second, any data and AI frameworks should be in compliance with existing government laws and regulations that address both data privacy, storage and security.

Third, AI decisioning frameworks must be used such that the AI is explainable in layman’s terms, transparent, fair, and human-centric (Singapore is playing a lead role in developing these standards).

Return to previous page