Facial recognition helps Indian authorities identify SIM fraudsters
Authorities in two Indian states have deactivated 17,000 SIM cards after using facial recognition technology to detect subscribers issued SIM cards against regulations.
The Department of Telecommunication (DoT) in the states of Bihar and Jharkhand found a total of 21,800 fraudulent SIM cards in April and May, The Print reports.
In order to prevent fraud, India limits the number of SIM cards to nine per subscriber. Some states have cut that number to six SIM cards per person. Scammers, however, have been using forged identity documents such as Aadhaar and headshots taken at different angles to register for multiple cards.
To prevent this, Indian authorities have come up with a new biometric tool called “AI and facial recognition-powered Solution for Telecom SIM Subscriber VeRification” (ASTR).
The news comes amid a rise in cyber crimes in eastern India. Fraudulent SIM cards are often used by fake call centers and in illegal betting and sex racketeering.
Earlier this month, DoT blocked 225,000 mobile SIMs in the states of Bihar and Jharkhand, according to the Times of India. In the release, the agency noted that the New Delhi DoT has carried out biometric facial analysis of more than 870 million SIM card subscribers in India.
“The analysis has been carried out using ASTR in collaboration with the Centre for Development of Advanced Computing. The result of analysis has been shared with Department of Telecommunication field units to ensure action against SIM fraudsters with the help of Telecom Service Providers and state police,” the statement reads.
The Indian government says it has used ASTR to block 835,000 SIM cards issued on fake IDs in the last three months. A half-million were blocked last year.
Privacy questions raised over facial recognition use
Although facial recognition tech is now being used across multiple states in India to combat fraud, questions about data privacy and ASTR are also coming to the fore.
India still doesn’t have a personal data protection regime or AI-specific regulation. But this hasn’t stopped the DoT from outlining plans to use ASTR to analyze the entire telecom subscriber base of India, The Hindu reports. This amounts to 1.17 billion subscribers.
The ASTR analyzes images of subscribers provided by telecom companies by grouping them into similar-looking images with the help of facial recognition technology. In the next step, it compares subscriber details and uses a string-matching concept called “fuzzy logic” to identify similar-looking names of users or other Know-Your-Customer (KYC information) to group them. In the last step, the system determines if the same face or person has acquired SIMs in in multiple names, dates of birth, bank accounts, address proofs, and other KYC information.
ASTR may be prone to inaccuracies, including false positives, if the facial recognition technology is faulty, the report notes. False identification with facial recognition has resulted in wrongful arrests and exclusion from social security schemes in India in the past.
Legal questions have also been raised about the use of ASTR. According to news organization Medianama, Indian citizens have largely been kept in the dark about the processing of their facial data. Medianama reported that the DoT failed to provide information about ASTR’s data safeguard and retention regulations and declined to release a copy of the final contract for the technology citing confidentiality.