Law in the Internet Society

Facial Recognition Technology and Mass Surveillance in India

-- By NinniSusanThomas - 22 Oct 2021

Background and Context

India has a long history of State surveillance - both physical and digital, targeted and bulk. In the colonial period (1857 - 1947), the British government regularly engaged in phone tapping, authorised by the 1885 Telegraph Act. Physical surveillance of entire groups and tribes was conducted under the 1870 Criminal Tribes Act. At the time of independence, much of the surveillance infrastructure created by the British was maintained intact; as judgments of the Supreme Court in the 1950s, 60s, and 70s show, warrantless searches and seizures of documents, phone tapping, as well as surveillance of movement of suspected criminals and “history-sheeters”, was a regular occurrence. Facial Recognition Technology is thus only the latest in a long history.

Why is this relevant? If it is explain why. If it is not, you need the space.

In 2019, National Crime Records Bureau issued a tender for the creation of a National Automated Facial Recognition System, much of which would be populated by gathering data from other privacy-invasive government projects, such as the Crime and Criminal Tracking and Network System (CCTNS). Even apart from a federal-level effort, FRT has been deployed in various states, and has been justified by the logic of policing: either ostensibly to guarantee the safety of women (as in Uttar Pradesh), broader public safety by apprehending people with a past criminal record before major public events (as in Tamil Nadu), or to guarantee safety inside school classrooms (as in Delhi). In addition, a Russian start-up has set up around 500 FR cameras across railway stations in Gujarat and Maharashtra, including Mumbai (through which 7 million passengers transit daily). According to a tracker set up by the Internet Freedom Foundation, there are currently 75 FRTS employed across the country, including major airports. Not just government bodies, but even state owned firms like NTPC have employed FRTS for attendance and consent is not required.

Why not just a few links to the sources from which you got this? A paragraph could become two sentences.

Constitutional Framework

In K.S. Puttaswamy v Union of India, a 2017 judgment handed down by a nine-judge bench of the Supreme Court of India, consistent jurisprudence of the previous four decades was consolidated and upheld, and it was declared that the Indian Constitution protects (an unwritten) right to privacy. The right to privacy was held to include decisional privacy, informational privacy, the privacy of the home and of documents, and the protection from arbitrary surveillance. This right could only be breached as long as the four-pronged test of proportionality (existence of a law, a rational aim, necessity, and proportionality stricto sensu had been satisfied). Arguably, the plurality opinion also held that mass surveillance would, per se, not be proportionate.

In addition, and in the context of FRT, the Indian Constitution recognises a right to equal protection of laws, and equality before law. Recent judgments of the Supreme Court have affirmed that the Constitution outlaws not just direct discrimination, but also indirect and intersectional discrimination.

FRT and the Constitution

It is clear that the use of FRTs violates both the right to privacy and the right to equality.

No, it is not clear. If you mean to make the argument, terrific, but it can't be assumed.

In a recent challenge before the Delhi High Court, it was pointed out that the accuracy of the FRT sought to be deployed is around 2%. This raises immediate rule-of-law concerns.

Under what circumstances. Lots of evidence collected by police is unreliable. Tips are far less than 2% reliable. And?

However, not only is accuracy a problem, but inaccuracy is also asymmetric in character: the quality of output is only as good as input data, and thus, FRT often reflects the prejudices of its designers. Studies worldwide have shown, for example, that FRTs are often more accurate in recognising white males, and make egregious errors in other cases.

Not in China, where because the same types of software are differently trained, white males are not the favored recognition subjects. And again, what is our ground of complaint because some evidence is unreliable. Need we not inquire into what is done with it, first?

In India, these inaccuracies map on to not just race and gender, but caste as well. The deployment of FRTs, thus, risks perpetuating existing structural injustices within the Indian criminal legal system, where there is already overrepresentation of groups such as Dalits and Muslims in prison.

Amit Shah told Parliament that their software wold recognize clothing. Might it not be time to decide whether there is any reality to the claims (either from GoI? or from its critics) in fact?

Furthermore, at the time of writing, there does not exist a data protection law in India. There is thus no regulatory framework in place that might mitigate any of these outcomes.

As though the legislation now moving, or any eventual legislation, will actually cover police agencies. See the pieces Mishi and I recently published on the subject in ET and ToI? , and the current ICR on the legislation.

This is particularly problematic when one realises that this data will also be used to extract particular data points such as the facial features and other biometrics, which the individual has not consented to sharing when entering a CCTV-surveilled zone, and these data points can be used to track future movements of the person. Therefore, integration of FRT with a network of CCTV cameras would make real time surveillance extremely easy.

Of course, that's the point. And? There's still no legal analysis to accompany the factual claims.

An absence of an effective data protection law also risks function creep. For example, it has been recorded that while the Delhi Police initially received permission to use FRT to track missing children, it was later used to profile and track individuals at anti-government peaceful protests. This is a classic example of function creep that violates the principle of purpose limitation. There is a violation here not only of the rights to privacy and freedom of association, but also an indirectly discriminatory effect on minorities, as many recent protests in India have been minority-led.

Finally, as the record around Aadhaar - India’s national biometric identification system - has shown, digital systems, when applied to welfare entitlements, can have a serious impact on socio-economic rights, especially because of the pervasiveness of false positives and false negatives. Making the provision of ration dependent upon facial recognition, for example, risks deprivation of the right to food (as has already happened with Aadhaar), especially in areas of India where access to digital networks is patchy and uneven.


The present and intended deployment of FRT in India - across a range of domains - raises serious questions of constitutionality. While some of these may be arguably addressed by a strong data protection law, others - such as deployment for surveillance or for welfare - will arguably fail the test of proportionality in all events.

This conclusion depends on the legal analysis not given. The best route to improvement is to reduce the space given to factual material that can be sourced and linked, so as to make room for the actual constitutional arguments you've kept in your pocket.

You are entitled to restrict access to your paper if you want to. But we all derive immense benefit from reading one another's work, and I hope you won't feel the need unless the subject matter is personal and its disclosure would be harmful or undesirable. To restrict access to your paper simply delete the "#" character on the next two lines:

Note: TWiki has strict formatting rules for preference declarations. Make sure you preserve the three spaces, asterisk, and extra space at the beginning of these lines. If you wish to give access to any other users simply add them to the comma separated ALLOWTOPICVIEW list.


Webs Webs

r2 - 05 Dec 2021 - 21:36:01 - EbenMoglen
This site is powered by the TWiki collaboration platform.
All material on this collaboration platform is the property of the contributing authors.
All material marked as authored by Eben Moglen is available under the license terms CC-BY-SA version 4.
Syndicate this site RSSATOM