Delhi and Kolkata’s police departments refused to furnish details about their respective facial recognition systems, in responses to RTI queries filed by the Internet Freedom Foundation (IFF), the digital rights advocacy body said on Tuesday. Additionally, the Telangana State Technology Services (TSTS), which was supposed to provide a facial recognition technology (FRT) solution to authenticate remote voters during the just-concluded Hyderabad municipal elections, told IFF that it had no intention to improve the present accuracy rate (80%) of the algorithm.
IFF asked the Delhi police if it had ever used the technology for investigations; whether any arrests had been made on the basis of results obtained from FRT systems; accuracy rates of such systems; information on databases in the department’s possession and on privacy impact assessments. However, the Delhi police told IFF that it could not share any of the required information, and referred to Section 8(d) of the RTI Act, 2005.
- IFF noted that Section (1)(d) of RTI Act contains provisions that allow the state to prohibit furnishing information on “commercial confidence, trade secrets, intellectual property”. The organisation said that the absence of information about privacy protections put in place by the Delhi police, there are concerns that the data collected from citizens might be accessed by Innefu Labs, the company that supplies the FRT technology.
To the Kolkata Police, IFF posed questions about the law that allows the use of FRT; whether it had sought legal opinion on usage; whether a privacy impact assessment was conducted; whether there are any guidelines or standard operating procedure; the specific purposes for using FRT; the cost of procuring and maintaining the systems. The Kolkata Police refused to furnish any information, citing Section 24(4) of the RTI Act, 2005.
- IFF noted that Section 24(4) allows “certain organisations” such as intelligence and security organisations exemption from disclosing information. “This means that the Kolkata Police is using facial recognition technology without there being any avenue through which transparency and accountability can be demanded from them for their use of such an invasive technology,” it said.
Meanwhile, IFF asked the Telangana State Technology Services (TSTS), which has supplied the State Election Commission (SEC) with an FRT system for voter authentication, for information on how the system works; what law authorises the use of FRT; accuracy of the algorithm from when FRT was first used by the SEC in the Kompally municipality elections; whether steps were taken to improve the accuracy; expenditure and so on.
MediaNama had reported earlier that the SEC had planned to deploy a remote e-voting system during the Greater Hyderabad Municipal Corporations (GHMC) elections. TSTS managing director GT Venkateswara Rao had then told MediaNama that remote voters would be authenticated using Telangana’s FRT-based Real Time Digital Authentication of Identity (RTDAI) system. This plan was subsequently dropped and replaced with an Aadhaar-based voting system. Eventually, the SEC dropped the remote voting idea altogether due to logistical issues.
- In its reply, the TSTS said that it had achieved an 80% accuracy while identifying voters. The remaining 20% of the cases failed due to quality of the photograph database. It said it had found no bugs in the system, and hence no steps were taken to improve the accuracy. TSTS also said the data collected by the FRT system was deleted in 14 days.
- Exclusive: Delhi Police working on arming PCR vans with facial recognition software
- Telangana plans to use blockchain-based e-voting, facial recognition, in Hyderabad civic polls
- Telangana State Election Commission drops e-voting plan for Hyderabad local body elections
- Tamil Nadu Police nab criminal using face recognition app, questions of safeguards and accountability remain
- Facial recognition, vehicle number plate identification: How Noida is ramping up street surveillance
- How Telangana is using the pandemic to push facial recognition tech on its population