How Aadhaar compromises privacy? And how to fix it?

The op-ed was published in the Hindu on March 31, 2017.

When assessing a technology, don't ask - “what use is it being put to today?”. Instead, ask “what use can it be put to tomorrow and by whom?”. The original noble intentions of the Aadhaar project will not constrain those in the future that want to take full advantage of its technological possibilities.  However, rather than frame the surveillance potential of Aadhaar in a negative tone as three problem statements - I will propose three modifications to the project that will reduce but not eliminate its surveillance potential.

Shift from biometrics to smart cards: In January 2011, the Centre for Internet and Society had written to the parliamentary finance committee that was reviewing what was then called the “National Identification Authority of India Bill 2010”. We provided nine reasons for the government to stop using biometrics and instead use an open smart card standard. Biometrics allows for identification of citizens even when they don't want to be identified. Even unconscious and dead citizens can be identified using biometrics. Smart cards, on the other hand, require pins and thus citizens' conscious cooperation during the identification process. Once you flush your smart cards down the toilet nobody can use them to identify you. Consent is baked into the design of the technology. If the UIDAI adopts smart cards, we can destroy the centralized database of biometrics just like the UK government did in 2010 under Theresa May's tenure as Home Secretary. This would completely eliminate the risk of foreign governments, criminals and terrorists using the biometric database to remotely, covertly and non-consensually identify Indians.

Destroy the authentication transaction database: The Aadhaar Authentication Regulations 2016 specifies that transaction data will be archived for five years after the date of the transaction. Even though the UIDAI claims that this is a zero knowledge database from the perspective of “reasons for authentication”, any big data expert will tell you that it is trivial to guess what is going on using the unique identifiers for the registered devices and time stamps that are used for authentication.  That is how they put Rajat Gupta and Raj Rajratnam in prison. There was nothing in the payload ie. voice recordings of the tapped telephone conversations – the conviction was based on meta-data. Smart cards based on open standards allow for decentralized authentication by multiple entities and therefore eliminate the need for a centralized transaction database.

Prohibit the use of Aadhaar number in other databases: We must, as a nation, get over our obsession with Know Your Customer [KYC] requirements. For example, for SIM cards there is no KYC requirement is most developed countries. Our insistence on KYC has only resulted in retardation of Internet adoption, a black market for ID documents and unnecessary wastage of resources by telecom companies. It has not prevented criminals and terrorists from using phones. Where we must absolutely have KYC for the purposes of security, elimination of ghosts and regulatory compliance – we must use a token issued by UIDAI instead of the Aadhaar number itself. This would make it harder for unauthorized parties to combine databases while at the same time, enabling law enforcement agencies to combine databases using the appropriate authorizations and infrastructure like NATGRID. The NATGRID, unlike Aadhaar, is not a centralized database. It is a standard and platform for the express assembly of sub-sets of up to 20 databases which is then accessed by up to 12 law enforcement and intelligence agencies.

To conclude, even as a surveillance project – Aadhaar is very poorly designed. The technology needs fixing today, the law can wait for tomorrow.