Вы понимаете? OSINT in Foreign Languages

It just takes one click in OSINT to land on a website in a foreign language. Investigations don’t have to stop here, if you have the right tools.

In today’s interconnected world, OSINT investigations lead us to foreign language content quite often. This does not mean we have to stop here. Thankfully, a broad variety of tools can support us in translating the content we find.

Before getting into specific tools, I have learned that you will receive the best results if you define the input language manually. Most tools can autodetect the input language, but if you’re working with short sentences or even single words, this might not function reliably. Sometimes translating very long sentences will also produce awkward results, splitting a long sentence into components could help in this case. That said, let’s have a look at some tools I use during my investigations.

First off, I would like to point out DeepL, a German company that trains AI to understand and translate texts. When it comes to translating content in German, English, Spanish, Portuguese, Italian, Dutch, Russian and Polish, DeepL has proven to be more accurate than other tools. You can copy and paste a text or upload a document to have it translated. I let the platform have a try at an excerpt from one of the older Keyfindings’ posts in German.

1

The next must-have is Google Translate. This extension should be installed in any browser to easily decipher pages on the fly. Next to translating complete webpages, it will show you the original text by hovering the mouse over that passage. In some cases this can be helpful, especially when Google tries to translate names of people, places or companies as well.

2.jpg

What if neither DeepL or the Google Translate extension work? Maybe you’re on a page that does not use the Latin alphabet, e.g. Chinese or Arabic, and some of the content is not ASCII-coded. This happens quite often when looking at Asian websites. Another case might be handwritten information in such languages. One of my favorite tools for this is on the Google Translate website itself. Next to the obvious copying and pasting of text, as well as uploading documents, Google allows you to use a foreign language virtual keyboards to input information.

3

However, this isn’t always helpful. In Arabic, letters vary in shape depending on their position in the word. This makes it hard for someone not proficient in Arabic to use the keyboard. Luckily, there is a workaround!

The Google Translate page allows you to draw what you see and based on that it will make suggestions and translate them. This works really well with any character-based writing, such as Chinese, Korean and Japanese, as well as with other languages that don’t use the Latin alphabet (Russian, Hindi, etc.). I have added a quick video to demonstrate how it works.

As an alternative, I looked into Windows Ink on the Microsoft Translator, but Microsoft currently doesn’t offer an Arabic handwriting package. However, it does offer Russian, Chinese, Hindi and several others character-based alphabets and languages.

When trying to translate subtitles in Videos, there is a workaround that was shared by Hugo Kamaan on Twitter, showing how you can use your cell phone camera to receive instant translations.

There are definitely more tools out there, so feel free to add anything you use frequently or that you think is missing in the comments.

Я надеюсь, что это было полезно для вашего расследования OSINT!

Matthias Wilson / 21.07.2019

How GDPR affects OSINT

The introduction of GDPR was a shock to many. While there are limitations, it doesn’t prohibit OSINT work completely. Find out what you can or cannot do when conducting investigations.

activity-board-game-connection-613508.jpg

In almost all OSINT activities, we process (e.g. collect, store, analyse, reproduce) personal data including names, addresses, user names, phone numbers, IP addresses and much more. However, the new data protection legislation introduced in the European Union in May 2018, the General Data Protection Regulation (GDPR), restricts the processing of personal data. Therefore, OSINT researchers need to have good understanding of how the GDPR applies to their situations, if only to stay on the legal side with their work.

In this blog post, we will discuss GDPR basics for OSINT researchers. We will not look at the exceptions. Processing personal data for household use or for journalistic purposes are for the most part exempted under the GDPR. Of course, the devil is in the details with respect to when these exemptions actually apply and this blog post will not go into those cases. Also, we will not look at OSINT for law enforcement use, as that has a different legal framework for dealing with personal data.

Hence, we will aim at OSINT in a commercial setting, where the researcher is dealing with matters such as background investigations, third party assessments and pre-employment screenings. Furthermore, this article will only discuss those GDPR aspects which are most relevant for OSINT work as we see it now. The GDPR is an extensive regulation with numerous aspects, which we cannot fully discuss in a single blog post.

Also please keep in mind: We are not lawyers and the implementation of the GDPR can differ between EU countries in numerous details. If you are reading this to seek legal advice, you should consult your friendly lawyer instead.

Summarising the GDPR, the aspects relevant for OSINT work are:

  1. You need a legal basis for processing personal data;
  2. You need to apply certain principles in the processing of personal data;
  3. The data subject of whom you process personal data has specific rights you need to understand, anticipate and honour.
  4. Understand if you are the data controller or the data processor.

Legal basis

Data protection regulation was never meant to render the processing of personal data impossible. Instead it is meant to balance the need for data exchange in our society on the one hand, and the fundamental right of privacy for citizens on the other. It is  therefore important to note that privacy is not an absolute right. The GDPR balances this right with other rights by restricting the processing of personal data to instances where there is a legal ground (Article 6). It lists six different grounds, of which three are potentially relevant for OSINT work: consent (Article 6a), legal obligation (Article 6c) and legitimate interest (Article 6f). We will discuss these three in detail.

Article 7 of the GDPR sums up the conditions for consent. Consent is tricky because the GDPR states that the data subject should have free choice when giving consent. However, for example, free choice in an employer/employee relationship does not really exist. Consent can – according to the GDPR – also be withdrawn at any time. So what happens when a data subject withdraws his or her consent halfway into an investigation? Or when it turns out afterwards, that the consent cannot be regarded as given freely?

Due to this ambiguous legal nature of consent under the GDPR, we believe that for OSINT investigations the use of consent should be avoided whenever possible. The risk that the data subject could afterwards argue that he or she had no choice than to give consent because of the consequences at risk, is simply too large. Moreover, often it is practically not possible to obtain consent especially if you, for example, are examining social media of the circles around your subject.

The second potential ground for processing personal data, that may be relevant for OSINT work, is legal obligation. This could be the case when your client has the obligation to identify their customers and the source of their funds under Anti Money Laundering (AML) regulations. Especially if you are instructed by a financial institution, this may likely be the overall legal basis for your OSINT work.

The third ground for processing personal data that may be relevant for OSINT work is a legitimate interest of your client.

What is legitimate interest? This is a tricky question, as it does not directly relate to any other law or regulation. Imagine the following scenario: A large stock-listed fashion company is in the process of hiring a new CEO. A final candidate is presented and the company puts him through a pre-employment screening. The company is known for their strong stance against animal cruelty and has supported many awareness campaigns in this regard. It has become their corporate identity. During the pre-employment screening pictures are found on social media, showing the CEO-candidate participating in annual fox hunts. If this information would leak when the CEO is already in position, it would surely cause a major scandal, possibly decreasing the stock value of the company and causing job layoffs.

The scenario described above could be a legitimate interest, as the financial situation of the company and thus the prosperity of many employees could be affected be the actions of one person. Nonetheless, pre-employment screenings are frowned upon in certain EU-countries.

Another example can be a simple fraud investigation, where you are tasked with identifying possible assets belonging to a person suspected of fraudulent actions. Your client claims to be defrauded and would like to initiate legal action. As such, the client has a legitimate interest to instruct you identifying possible assets of the perpetrator.

In sum, as an OSINT investigator, you should always understand and document the legal basis to process personal data before conducting your research. In a commercial setting, that will most often be either be a legal obligation or a legitimate interest of your client. We advise to always properly document the legal basis, for example, by explicitly detailing the situation in an engagement letter or contract for the work. More in general, and from an ethical point of view, we believe that you should always want to understand what the purpose of your work is and the interest of the client.

Principles

Regardless the exact legal basis, Article 5 of the GDPR imposes a number of principles for the actual processing of personal data. Each of these are relevant for OSINT work and therefore we will discuss them all.

  1. Lawfulness, fairness and transparency

Lawfulness is self-explanatory: You need a legal basis to process the personal data. This should be one of the six legal bases as provided in Article 6. Furthermore, you should not hack, steal or lie to get the data and to prove this, you need to document the sources – which a professional OSINT researcher would do anyway. Not only are these actions unfair and unethical, they may as well be illegal in many cases. Fairness also relates to proportionality. Is the amount of personal data you collect on the data subject proportional to the task at hand? Transparency is touched upon in more detail in Article 14 of the GDPR (situations where the data is not collected from the data subject), we will discuss this further along.

  1. Purpose limitation

Again, pretty easy. What you collect for one purpose, shouldn’t be used for another incompatible purpose.

  1. Data minimization

You should minimise the amount of personal data to what you really need. As much as necessary, as little as possible. So when you scrape gigabytes of data, only the information relevant to your task should be retained.

  1. Accuracy

You have the obligation to make sure that data is of good quality, thus do not use outdated information or data of which you know it is incorrect. Especially when working with people search engines, you will stumble upon a lot of old or false info (false flags) on individuals. You are responsible for verifying the data where possible before further processing or reporting.

  1. Storage limitation

How long do you retain your records? The GDPR states that it should not be longer than needed. In an ongoing litigation that could be years, but often not more that general data retention obligations such as in civil or tax law should be followed. Again, this can differ per country.

  1. Integrity and confidentiality

You are responsible for keeping the personal data you process secure. The fact that you may have collected it from open sources is completely irrelevant, it shouldn’t be publically disclosed from your side. The adagio here is: If you cannot protect it, don’t collect it.

  1. Accountability

The basically means that when processing data you should not only adhere to the previous principles, you should also be able to demonstrate afterwards that you did. The GDPR has shifted the burden of proof to those processing personal data, so you need to document!

A most efficient solution to document on how you comply with the GDPR is to draft an investigation protocol in which you describe how you process personal data and how you apply these principles to your work. In any assignment or when you get questions from the Data Protection Authority (DPA), you can refer to the protocol.

Data subject rights

The third category of GDPR provisions relevant for OSINT, is the data subject rights. According to GDPR, a data subject has rights of:

  • Notification

Every data subject shouldbe notified which personal data is processed by whom, why and where. GDPR Article 13 and 14 state that no matter if collected from the subject itself or without knowledge of the data subject, they must be informed in order to be able to contest this data collection and processing. Again, this proves to be tricky, especially when conducting investigations against a certain subject.

In some cases, the notification might be disproportional in regards to effort that has to be made in order to inform the data subject.

  • Every data subject has the right of access, right to rectification, right to erasure, right to restriction

If informed, a data subject has the right of access to all data stored on him- or herself, the right to restrict further dissemination (if not necessary by law), and last but not least the right to have all data deleted. Of course, this must be weighed against the legal basis or legitimate interests upon which the investigations took place.

There is also an exemption on these rights possible under article 23 of the GDPR which states that Member States can limit the obligations and rights under the GDPR in certain instances. This has to be done by law and every Member State may have implemented this provision differently. The Netherlands have Article 41 of their national implementing law which fully integrates Article 23 GDPR.  Does your country have the same?

Of course, if you would use an exemption, you need to document the circumstances and considerations on why you think that this exemption is justified. Be careful with this and understand the local implementation of the (conditions for) exemptions before you apply them.

Are you the controller?

A final important point relevant for OSINT work is whether you are the data ‘controller’ (the one who determines the purposes and means of the processing of the personal data) or the data ‘processor’ (the one who processes personal data on behalf of a controller).

The easiest situation is where you are the data processor and you process data under responsibility of the data controller. In those instances, most of the legal obligations are  mainly the responsibility of the data controller, which usually is your client.

However, the determination on whether you are the processer depends on the level of freedom you have in choosing the purpose and methods of processing the data. If you determine the purpose, types of data and methods applied, you cannot argue that you are just the processor, as you in fact ‘control’ the data processing.

Having a data processing agreement with a client – or adding a section on data processing to your existing agreement – is an important prerequisite to be regarded as the processor. To be regarded as the processor, the agreement should clearly show that you are instructed for a specific purpose, looking at specific types of data and applying specific methods (and excluding anything else) and reporting in a specific way. Once more, consult your friendly lawyer for more details.

There is limited jurisprudence on GDPR issues and this means, that in many instances it is not exactly sure how the GDPR will be interpreted exactly. The principles should give some guidance, nonetheless, in high profile cases make sure you discuss these matters with your client and (their) inhouse legal counsel.

We have given a general overview of the most relevant aspects of the GDPR for OSINT work. However, we realise that we are not complete in that matter. The GDPR has a number of other relevant articles, for example on processing of special categories of personal data, but there are limits to what we can cover in one blog post.

So, get your copy of the GDPR today as well as a copy of the implementing law in your country, read it, seek advice, understand it and most important: Comply with it!

Ludo Block & Matthias Wilson (I just chipped in a little)/ 11.06.2019

Tracking a Hacker with OSINT

My blog has been hacked! Someone defaced the page and looking into the technical details didn’t provide any leads to the culprit. Maybe OSINT can help in this case.

1

Today’s article will look into cyber attribution and how OSINT can help identify the perpetrator of a cyberattack or other hacking exploits. Keep in mind, as long as the perpetrator does not make any mistakes it will be hard to track him down. Even if the actual person behind an attack cannot be found, hints on the hacker’s background may help narrow things down to a specific target group or origin. Let us have a closer look at the defacement shown above.

As stated, looking into technical details (IP-address, code, etc.) did not reveal anything useful. So we have to take a closer look at the tag and handle that was placed on our site. A reverse image search was conducted and did not show any results. The hacker goes by the name “drag0nw1ng١٩٨١”, this exact search-term also came up inconclusive. The Arabic numbers in the handle may be an indicator for the hacker’s cultural background. Next up, we will search for the handle in different variations, including a “standardized” one:

2

Not many results to look at here, so we can easily go through each and every page. Next to a Russian PlayStation profile named Dragonwing1981, we stumble upon some interesting results that might be related to our case.

3

Several data-breaches and leaks show an email address using the exact name. Dragonwing1981@yahoo.com was registered to a member of an internet forum called Kataib Hezbollah. This forum in Arabic language no longer exists and was used to disseminate terrorist propaganda. Since our hacker used Arabic numbers in his handle and the handle seems quite unique (based on the low amount of Google results), the email address might be linked to our guy.

The oldest mentioning of “dragonwing1981” came from another internet forum. In August 2004, the forum was hacked by someone with the email address we found before:

4

Research done by the forum members linked the perpetrator to Iraq:

5

Looks like things are coming together. There is one more approach we can try, in order to back our claims further. When using the password reset function in Yahoo, it gives you parts of the phone number (without the country code). Let us see what happens, when we try to reset Dragonwing’s password:

6

07 is the operator code used by Iraqi mobile networks and the length of the number also fits Iraqi mobile phone numbers. Luckily, Yahoo (unlike Google) displays the exact amount of digits of a phone number.

Let us review the evidence we have collected so far:

  • Use of Arabic numbers in the handle
  • Unique handle, not found often on the internet
  • Username and a related email address found in an Arabic internet forum
  • Email address used in a hack in 2004, identified as possibly originating from Iraq
  • Phone number linked to the email address possibly an Iraqi mobile phone number

Can we be sure that all these pieces of evidence are really linked to each other? Not really, but that is why we use words of estimative probability in intelligence analysis. Cyber attribution is not always about tradecraft, infrastructure or the malware/attack itself. Digging into individual actors may help shed light upon the origins of cyber-attacks and the OSINT process shown above should always be incorporated into any research effort as soon as “personal data” (e.g. tags, names, handles) is involved.

Of course, we could just send Dragonwing1981 an email and congratulate him on his defacement. However, unlike other stories on my blog, this one is completely made up and is based on a CTF-task I created for the OSINT courses I instruct. As far as I am concerned, Dragonwing1981 is innocent…

Matthias Wilson / 02.05.2019

The Nigerian Prince from South Africa

Great, another Nigerian prince in your inbox. Instead of deleting it, why not answer for a change. I did and it turned out to be quite interesting.

Last week, I received my first Nigerian prince scam mail (also known as 419-scam) in German. I assume someone put a lot of work into this, so I thought I would answer. Although the message was apparently sent from jefaturaestudiositurbi@valencia.es, I was to reply to wong.shiu@accountant.com. This email supposedly belonged to Mr. Wong, the banker who was handling the case.

Let us have a look at the message header first, before answering.

1

Even if I would have answered to jefaturaestudiositurbi@valencia.es, the email would have been sent to wong.shiu@accountant.com. I assume the email was not actually sent from the @valencia.es domain in the first place and that this was just used to bypass my spam filter. Next up, I wanted to see if I could find any leads to where the email was sent from.

2

The initial ‘Received’ entry in the message header points to a South African IP-address belonging to a mobile provider. It also appears to have been sent through a Huawei 3G/4G WiFi router.

Next up, I set up a new Gmail account to communicate with this Nigerian (South African) prince. Sure enough, I received an answer within minutes. The reply contained additional information regarding the deal and was clearly a very bad Google translation of an English text. Again, this message was sent from the same IP-address. We emailed back and forth several times until I was asked to provide some ID, an address and a phone number. So I did.

3.jpg

Apparently, Mr. Wong thought this was funny as well. For the first time I actually received a response that was not just copied and pasted from a pretext.

3-1

“You dey gather my fmt” – This actually translates to: So, you are one of those guys that collect my pretext. At this point, Mr. Wong also started using a different email to communicate with me: wong.shiu@mail.com. Again, I checked each message header. While several different IPs were used, they all belonged to South African mobile providers.

4

The conversation went on for quite a while and I was surprised that Mr. Wong kept answering.

5

The following day I received another scam mail that looked just like to first one. The only difference  was that the name of the banker had changed (and thus the reply email) and the promised sum of money was a lot higher than in the first email. It sure looked like this was also the work of my friend Mr. Wong, so I decided to answer to this new email as well.

6

Unfortunately, Mr. Wong did not answer any more. Looking into all the emails again, I could clearly see a pattern. Each IP-address could be traced to South African mobile providers and all emails were sent through Huawei 3G/4G WiFi routers. The language used also hinted towards Africa in general. Furthermore, over the course of two days I noticed that Mr. Wong began answering around 09:30 (CET), leading to the conclusion that he must have been in the same time zone (or nearby) if this was his 9 to 5 job.

If you ever try this yourself, please make sure to use a clean email address and do not download or open attachments. If you keep this in mind, you might have some fun with a Nigerian prince yourself. As for Mr. Wong:

Mr. Wong,

If you ever read this, feel free to contact me again. I can’t promise I’ll pay the advance fee you requested, but I’m always there for you if you need someone to chat.

Yours sincerely,

Matthias Wilson / 19.03.2019

Building a Hells Angels Database with Hunchly

Today I will teach you about Hells Angels and Hunchly and how one of these two is useful when looking into the other.

In the past year, I have worked two cases in which I stumbled upon links to Hells Angels while investigating individuals. I was surprised how much information people affiliated with this group shared publically on Facebook and other social media sites. Whether they were just supporters or full members, it became quite clear that they did not care about data privacy. Most profiles had open friend lists, some of them displaying thousands of friends. Hells Angels affiliates are not hard to find. You will likely stumble across one of the following acronyms and/or terms on their profiles: AFFA (Angels forever, forever angels), HAMC (Hells Angels Motorcycle Club), Support 81 (8 = H, 1 = A), SYL81 (Support your local Hells Angels), Eightyone.

There are a couple more, but this article is not about the Hells Angels per se. Since these individuals have so much open information on Facebook, their profiles are the perfect playground to try out Michael Bazzel’s Facebook tool on IntelTechniques.

I had just finished working on the first case and subsequently erased all the data linked to that case, when a second case soon revealed links to Hells Angels as well. If only I had saved some data from my first case. I roughly knew where I could start off, but most of this knowledge came off the top of my head and was sketchy. Before I started the second investigation, I made sure I wouldn’t make the same mistake again and decided to use Hunchly to save my findings. That way, if a third case with the same links should ever occur, I will have a great starting point. For those of you who do not know, Hunchly is a web capture tool. It automatically collects and documents every web page you visit. The best part is that it indexes everything, so you can search within the data afterwards. Using this amazing tool allowed me to create a fully searchable Hells Angels database!

First off, I created a new casefile and then let Hunchly collect Facebook friends lists of people affiliated with my target or any Hells Angels in the area my target originated from. As some of the profiles had thousands of friends, I used a little Chrome extension (Simple Auto Scroll) to automatically scroll down friends lists, so they would be captured in whole. Whenever I looked at profiles and found information that could not be automatically indexed, I would take notes in Hunchly or tag (caption) pictures. I have learned that a lot of intelligence can be obtained by closely looking at pictures on social media. In the following example, one Hells Angels member had obscured the tags on his vest. Based on the information in his profile, it became clear that he must belong to the Aarhus chapter in Denmark. I tagged this picture, meaning it would pop up if I ever searched for “Aarhus” in Hunchly.

1

I ended up tagging all pictures that included chapter names, functions, nicknames or general indications on the location. If I am interested in finding the security chiefs and weapons masters, all I have to do now is search for “Sergeant at Arms” or known abbreviations. Looking for “arms” gives me several results in Hunchly.

2

The first two are displayed because I manually tagged these pictures and added a caption. The third result is from a webpage that Hunchly captured, in which the person actually listed “SGT At Arms” as his current occupation. Hunchly also allows you to refine searches. I can narrow these results down and, for example, only search for Sergeants at Arms in a specific chapter. Searching for “arms + sacramento” only reveals one result, which I had captioned with the information I saw in the picture. As you see, the picture is actually mirrored.

3

All collected data is saved offline. Should the online profile ever change, be locked down or deleted, I still have a version to work with. By using Hunchly and remembering to tag pictures with captions and also take notes on webpages, I have created a useful database on Hells Angels Facebook profiles. From here on, it is also always possible to go to the live versions of webpages, so any updates can also be captured within the same casefile.

If you are not using Hunchly yet, I suggest you have a look at it. The use case described above is just one of many. Furthermore, if you ever come across friendship requests from people named “AFFA” or “HAMC”, you might want to think twice before accepting them. Or else you might wind up in my Hells Angels database.

Matthias Wilson / 07.03.2019

The World’s Best Sock Puppet…Not!

There are lots of great guides on how to create sock puppets. Rather than showing you a good example on how to do so, this post shows a horrible example that has been used in a recent phishing attempt.

I received a request to connect on LinkedIn from what clearly is coming from a badly created sock puppet. This request is actually a cheap phishing attempt, aimed at getting a hold of my phone number. Basically, the perpetrator made every mistake in the book when creating the profile. Let me walk you through the red flags I encountered. Or: How not to create a sock puppet!

Red Flag 1:

Bad English. Have a look at the message I received.

1

When looking at the vita, it is clear that Liya Lei should have better English skills!

Red Flag 2:

No contacts (blue box). As you can see, the profile has no listed number of contacts. This is an indicator that it was just recently created or that it is not well-tended.

Red Flag 3:

UKTI does not exist anymore (red box). UKTI stands for UK Trade & Investment, a UK government department working with businesses based in the UK. In July 2016, UKTI was replaced by the Department for International Trade. Again, either this is just a bad sock puppet or an account that is not well-tended. In both cases, it does not seem trustworthy enough to hand over my phone number to.

2

There are some additional steps that can be conducted to verify accounts. The first step is, of course, running the name through Google. In our case, it did not produce any results directly linked to the person shown in the picture. Furthermore, a reverse image search should be performed as well. Forget Google, use Yandex for this. Unfortunately, neither Yandex nor Google were able to find the picture.

Another method to verify LinkedIn accounts, is searching for the person’s email. Assuming the account is real, we should be able to identify a company email address. A quick Google query reveals that the domain ukti-invest.com was among those used by said organization. Next up, run the domain through hunter.io to gain information on the pattern used for their email addresses.

3

Ukti-invest.com uses “firstname.lastname”, so we can now check if an email address belonging to Liya Lei exists. I checked the email address on verifyemailaddress.org and it clearly shows that while the domain exists, the email address we provided does not.

4

I also tried a couple variations, including different domains, such as gov.uk, as well as other naming patterns just to be sure.

Following these steps, I have pretty much proven that Liya Lei’s account is a total hoax. A very bad sock puppet set up to phish my phone number. A final note to whomever tried to fool me:

Dear Sir or Madam,

Next time try harder! There are plenty of guides out there on how to build a credible sock puppet. Your cheap attempt is actually quite insulting and did not even push my OSINT skills to a limit.

Yours sincerely

Matthias Wilson / 21.01.2018