- Part 1 - The Weak Security Of The Portuguese Government’s Authentication System
- Part 2 - Chave Móvel Digital Multiple XSS Vulnerabilites
- Part 3 - Chave Móvel Digital Phone Number Leakage
- Part 4 - Chave Móvel Digital Log Out Not Working
Back in January 2018, after about half an hour of exploring the “secure” authentication system provided by the Portuguese government I found various security issues and vulnerabilities. I will briefly explore them in a series of blog posts.
In this first blog post I will explain what is
Chave Móvel Digital, as well as the sensitive information that they
are supposed to be restricting the access too. I will also explore the
overall weak security that the system has.
This whole series of blog posts is a result of my findings in a short period of time. It’s a compilation of the notes that I took, as a draft to be submitted to responsible agency, as part of responsible disclosure. A more thorough security audit is likely to reveal a lot more problems. If about 30 minutes is all to took to find those problems, what more would a malicious, motivated attacker would find.
I’m hoping that the issues will be fixed as soon as possible.
On January 21st I emailed
email@example.com (email of the responsible agency),
describing the vulnerabilities and security issues, as well as offering
a more detailed description draft and Proof-Of-Concept code, but
obtained no answer.
What Is Autenticacao.gov.pt and Chave Movel Digital (CMD)?
Autenticacao.gov.pt’s homepage has the following text:
O sítio oficial dos meios de identificação eletrónica,
assinatura digital e autenticação segura do Estado.
which translates to:
The official site of the means of electronic identification, digital signature
and state's secure authentication.
Basically, it’s an authentication system that allows the Portuguese citizens to authenticate to administrative public services. Once authenticated, you can, among other things:
- register a birth of a child
- register a vehicle
- register a company
- request a birth certificate
- obtain the criminal record of an individual or a company
- submit/consult revenue of an individual or a company
- sign documents
New functionalities are constantly added. As you can see, it is a very useful system, since it allows you to save a lot of time. It allows you to do many things from the comfort of your home and in minutes, instead of spending hours at the pubic administrative services.
Despite being very useful, it does expose some sensitive personal information and allows you to do some sensitive actions. If this gets into the wrong hands, it can lead to many problems, such as identity theft.
For this reason, it is very important to have a secure authentication system. This is what Chave Móvel Digital (CMD), which translates to Digital Mobile Key, supposedly does. The main problem with it is that its security is questionable.
Bad Security From The Ground Up
So now let’s briefly talk about the security of the Chave Móvel Digital authentication system. In December of 2017 me and my girlfriend went to a Christmas fair. Among other things, there were various company booths. One of them was VR-related, so we decided to go in and check it out. After playing around with the Microsoft’s HoloLens for a little bit, we were approached by a lady there that asked us if we have already activated the Chave Móvel Digital system. After explaining to me what that was, she told me that it soon would be mandatory to have it activated. I could either do it there now or go to the public administrative organs, wait in queue for hours and do it there.
Motivated by the interest in the security of this new system and the potential future time savings (bureaucracy is a big issue in Portugal), I decided to activate it there.
The Authentication Process
Let’s look at the authentication process. To login using the Chave Móvel Digital system you have to enter your mobile phone number and your PIN. If the entered details are correct, a code is sent to your mobile number, which you have to enter to login.
The issues here are the following:
- PIN can only be 4-8 numeric digits
- SMS as the only second factor authentication option
The first one is the most serious one. PIN is essentially your password. The issue here is that it only be numeric its length is limited to be between 4 and 8 characters.
Let’s look at the numbers.
8 digit numeric-only passwords means that
10^8 = 100000000 possible combinations. This is nothing for
modern computer. For example, let’s see how long will it take to generate
all of the possible combinations on my machine. Assuming that the file
gen.py contains the following code:
Here is what
time python gen.py gives us:
$ time python gen.py python gen.py 3.20s user 0.01s system 99% cpu 3.208 total
While this is not the most accurate way of measuring the process execution
time, it’s good enough to transmit the idea of how quick modern computers are
at generating the entire key space. This is to give you an intuition of
how small the number of possible PINs is. It’s
2018 and the Portuguese government
seems to think that 4 to 8 digit passwords are secure.
Let’s compare this to an
8 character long password that allows
numbers (10 different:
0-9), letters (52 different:
A-Z) and special characters (32 different). For every position we would have
10 + 52 + 32 = 94 possible values. This means that there would be a total of
94^8=6095689385000000 possible combinations. If we use the rule of three
to estimate how long it would take my computer to generate the entire key
space, we would see that it would come out to more than 6 years.
As an interesting note, when I was signing up with the system, the lady
told me that from her experience people usually choose
4 digit PINs, so
that they’re the same as their phone PINs. This seems like a reasonable
assumption to take. In this case, the simple task is made even simpler, since
the number of possible combinations is lowered to
10^4 = 10000:
Here is what
time python gen.py gives us:
$ time python gen.py python gen.py 0.01s user 0.01s system 98% cpu 0.015 total
I presented the time measurements just to give you an idea of how
small the possible key space is. Of course, we can’t conclude that the PIN
will be bruteforced in
3 seconds. The system limits the number of
attempts that you can perform. However, if you think that this makes bruteforcing
impossible, you’re wrong. I will now demonstrate why.
Let’s assume that the number of login attempts is limited to 3 (in reality,
autenticacao.gov.pt allows more than 3 attempts). Let’s also assume that a lot of people will in fact
4 digit long PIN. This means that for each individual user there will
3/10000 chance of guessing his PIN. Pretty unlikely, right?
Well, this number also means that for every 10000 accounts the attacker
will guess the PIN of 3 of them.
To make the things worse, the numbers above are assuming a totally random
distribution of PINs, which is not true. A data analysis of almost 3.4 million four digit passwords showed that
make up almost
20% of used passwords. Considering this, the probability
of guessing the PIN of a single account would be about
Therefore, for every 10000 accounts the attacker would guess the PIN of 2000 of them.
Why doesn’t the system allow passwords that contain numbers, letters and special characters? Maybe it comes down to a legacy issue, but it is definitely possible to write middleware that would take care of that. The system is supposed to protect very sensitive data, after all.
As an example, let me present to you a very simple solution that would make the existing system more secure. It would do it so by allowing the users to use a password that consists of numbers, letters and special characters, while keeping the core legacy backend unchanged. I’m going to take the following assumptions:
- the backend requires a 4-8 digit
- all of the PINs are stored in plain text
A very simple solution would be to do the following:
- When the user is registering/changing his password: hash the password provided by the user with the phone number and store it in the database. Generate an 8 digit PIN and store it. This PIN will never be revealed to the user.
- When the users tries to log in: hash the provided password with the phone
number and compare this result to the one that you have stored in the
- if they are equal, send an “OK” to the backend, which will then obtain the PIN from the database and use it as needed
- if they are not equal, fail the authentication
I am in no way suggesting that this is the best solution, because it is not. It is, however, a very simple one, that would require very little changes to the overall system and keeping the core legacy backend unchanged. All you would have to do is add a new table to the database (this way you will keep your existing tables unchanged) for the password hash and change the authentication code in the middlware between the client and the backend system. Even if the system is different from the one in my assumptions, you can adopt the solution with relative ease.
The second point relates to only allowing SMS as the second-factor authentication. While being a lot better than not using two-factor authentication at all, it’s not the most secure option. Let me give you two examples. By using some social engineering, if an attacker can get access to your personal information, they can contact your phone company and move your phone number to a new SIM. By exploiting the vulnerabilities in the SS7 protocol the attacker can read all of your text messages.
Allowing the use of other authentication options, such as Google Authenticator would make the second-factor authentication a lot more secure. In general, I have a feeling that the engineers assumed that there was no need for strong passwords, since two-factor authentication is required and only the user has the access to his mobile phone number. This assumption is, of course, incorrect. While it is true, that if you’re using a good second-factor authentication, it’s safe to weaken your password, this does not apply to making your password a trivial 4-8 digit one.
Now, this just smelled bad security. I was certain that I could more problems. I was correct. The security issues that I’m going to be describing in this series of blog posts were found after about 30 minutes of exploring.
There is now an option to receive your second-factor code as a private message on Twitter or be sent to your e-mail address. Those are not enabled by default and you have to do it yourself. I am not sure when those options were added. In any case, when logging in into your “Personal Area”, you only have the option of using your mobile phone number.
In this first post of the four-part series I described why the security of the authentication system is flawed. This is also what motivated me to explore further: I was sure that it wouldn’t be hard to find more issues, since the “security” of this “secure authentication system” did seem like an afterthought.
In the next blog post, I will be describing the multiple XSS vulnerabilities present in the authentication page.