10 things you should know to get your national ID right
At an Internet Society Asia-Pacific bureau meeting on privacy last week, a representative of a government asked about how we can have national ID systems that protect privacy. From what I gathered from conversations that followed, several governments are looking to set up national IDs in the Asia-Pacific region.
While having a national ID system is, by itself, problematic, here’s a quick list I made for how not to screw up the national ID if you want to have one despite its risks:
1) Make it optional
A mandatory national ID is a recipe for surveillance and runs the risk of citizens’ data being compromised in one way or the other. Even an optional national ID stands the chance of becoming “voluntary but mandatory” – as the joke about Aadhaar goes – where making it mandatory for services that cover almost the entire population, such as getting mobile services means that it becomes mandatory for the entire population. Remember that data will be collected, stored, shared and compromised.
By making it mandatory, you rob people of the choice of not getting a national ID, and thus rob them of the option of protecting themselves against potential hacks, leaks and malafide intent, and persecution from dictators.
National IDs and associated data do get hacked and leaked. Estonia, the poster-child of digital governance, has had to suspend its digital ID cards. Spain is facing similar issues. Meanwhile, 143 million social security numbers were compromised in the US and at least 130 million Aadhaar numbers were published online by the Indian government.
2) Make it one of the many IDs for authentication
Federated means of identification ensures that people can identify themselves where needed without necessarily compromising the only ID they have. For example, credit card theft doesn’t affect debit card usage and the theft of a driver’s licence as an ID doesn’t affect the collection of bank subsidy.
However, the more linkages you create for a single ID and the more places people use it, the risk of identity theft increases. By limiting usage – for example, for bank accounts, mobile phones, college exams, mutual funds, stock market trading to a single ID, you run the risk of making that national ID a single point of failure for an individual. Databases will get compromised. Thus, you also run the risk of making it a single point of failure for your entire citizenry/population.
3) Give control to users to change and revoke an ID
Every instance of usage should be shared with the user who is supposedly using the ID, just like with messages and cash withdrawal. This helps because in case the ID is compromised, users can then contact the ID authority or the data controller, and ask them to revoke or freeze usage. The most important aspect is that the ID number must not be permanent and non-changeable.
The Indian passport, for example, once stolen, is re-issued with a different number. There is also the issue of bounded rationality: that people don’t necessarily fully understand the implications of what they are signing up for.
Thus, if they feel, say a few years later, that having a national ID puts them at the risk of their data getting misused or compromised, they must have the right to revoke it. Consent should not be forever.
4) Enforce usage of derived authentication/pseudonymisation
The usage of derived identification numbers or of artificial and/or temporary identification numbers means that the core national ID does not typically get exposed. This means that each derived ID has a limited use case and/or a limited shelf life, and this mitigates the potential harm from a single ID leaking or being exposed.
A national ID by itself should never be a means of identification. For example, see what Austria has done.
5) Give citizens the legal right to recourse
Legal recourse is a deterrent against misuse. While it may sound inexplicable that someone cannot sue an entity that has stolen their data, or sue a data controller (which holds data) against improper storage/security or conduct when it comes to processing or storage of their data, this has happened in case of India’s national ID project, Aadhaar.
There is no legal deterrent against, say, publishing data online, which has been done by 210 government websites, and just four of which have led to the publishing of data for 130 million. The option of a legal recourse against something that compromises your personal data acts as a deterrent against this. All it takes is for one case to make everyone change the way they operate.
6) Purpose limitation for national ID usage
A national ID that is digitally linked to, and can authenticate a large number of services, is likely to be seen as a key reason, and a significant convenience, for having a national ID.
However, it’s important to not link the national ID for things where they are not necessary, where you don’t have an option for something to function without the linkage of that ID.
The more the use cases for the national IDs, the more the risks of social hacks that can compromise even the most digitally-literate citizens. This leaves the illiterate and the digitally illiterate or neophytes even more vulnerable: they do not know the risks of the consent that is given.
This is where consent is insufficient.
Most importantly, the national ID should not be linked to sensitive personal data, such as DNA banks, health records, et al.
The national ID becomes especially problematic when it is linked to external, non-governmental databases such as mobile numbers, and used to share personal data with a mobile operator, given that governments, ID authorities do not necessarily have the wherewithal or capacity to monitor the security practices of third parties.
7) No biometric authentication
I can’t emphasise this enough. Biometric information is a permanent identifier, and can be easily compromised. Fingerprints can be copied from high-resolution photographs, or from that glass that you just held. So can the iris. Social hacks can lead to copying of fingerprints, say, if someone puts a fake authentication machine before you, before they place a real one. Sure, credit cards can be copied too, but cards can be replaced. Your fingerprints cannot.
If you have a permanent ID (say, Aadhaar) and a permanent password (your fingerprint), one getting compromised means someone only needs the other factor, and you are compromised forever. Note that mobile One Time Password isn’t secure either, and has been used in hacks in the past, and mobile networks operate on a maximum of 44 bit encryption.
Outside of security, also note that digital, biometric authentication suffers from other issues: for example, lack of internet connectivity for authentication, fingerprints getting worn out – an issue for manual labourers and the aged. Such situations could end up depriving those who really need it for accessing their benefits.
8) Data protection law comes before national ID
One of the key mistakes that India made with its national ID (Aadhaar) was that a data protection law isn’t there yet, but the national ID has been around for almost seven years. An act governing the national ID wasn’t even passed until almost five years of the ID being around. Thus, no privacy principles have been established, and no norms regarding data collection, storage, transfer, linkages, sharing and disposal are in place.
There are no penalties in place for violations of these norms either. It’s a free-for-all. Do not do this.
9) Don’t hurry, don’t push for 100 per cent penetration
Undue haste and the creation of deadlines for enrolment for a national ID can create panic among citizens, and such situations lend themselves to exploitation and fraud, especially in scenarios where people are being denied their entitlements, or run the risk of key accounts – like their bank accounts – getting frozen for lack of having a mandatorily linked ID. Don’t subvert democracy for increasing the speed of enrolment.
Instead, if you must have a national ID, roll it out without undue haste, at people’s convenience, with improved checks and balances. Above all, don’t be daft enough to outsource enrolment to third party agencies, paying them on a per enrolment basis, which then creates a perverse incentive of maximising enrolments.
Speed causes more harm than good here.
10) A budget for citizen awareness, education and grievance redressal
Something as significant as a national ID project can lend itself to a lot of misinformation and misinterpretation. There are also likely to be several issues related to enrolment and registration, as well as authentication.
This is, of course, besides the point that there are excellent reasons for not having a national ID:
a) Linking multiple databases to a single ID is harmful for citizens, and puts them at risk. It is more likely to form the basis of a mass surveillance system, and has a risk that a fascist regime can use it for ethnic cleansing or segregation.
b) It doesn’t address terrorism or volume-based pilferage of benefits, which are likely to continue despite a national ID. I can, in fact, be used to deny people benefits.
c) It also creates a new power centre, from the perspective of a single body which has the power to delist an individual from the database, thereby delinking them from essential services (if those are linked to a national ID).
d) It’s also worth nothing that data is a toxic asset, and the harms of losing data when it leaks or gets hacked is far greater than the benefit of collecting and storing that data.
(This article was first published in MediaNama.)