Beyond the hash: How unsupervised machine learning unlocks the true power of JA3

Max Heinemeyer, Director of Threat Hunting | Friday June 22, 2018

Introducing JA3

JA3 is a methodology for fingerprinting Transport Layer Security applications. It was first posted on GitHub in June 2017 and is the work of Salesforce researchers John Althouse, Jeff Atkinson, and Josh Atkins. The JA3 TLS/SSL fingerprints created can overlap between applications but are still a great Indicator of Comprise (IoC). Fingerprinting is achieved by creating a hash of 5 decimal fields of the Client Hello message that is sent in the initial stages of an TLS/SSL session.

JA3 is an interesting approach to the increasing usage of encryption in networks. There is also a clear uptick in cyber-attacks using encrypted command and control (C2) channels – such as HTTPS – for malware communication.

The benefits of JA3 for enhancing rules-and-signatures security

These near-unique fingerprints can be used to enhance traditional cyber security approaches such as whitelisting, blacklisting, and searching for IoCs.

Let’s take the following JA3 hash for example: 3e860202fc555b939e83e7a7ab518c38. According to one of the public lists that maps JA3s to applications, this JA3 hash is associated with the ‘hola_svc’ application. This is the infamous Hola VPN solution that is non-compliant in most enterprise networks. On the other hand, the following hash is associated with the popular messenger software Slack: a5aa6e939e4770e3b8ac38ce414fd0d5. Traditional cyber security tools can use these hashes like traditional signatures to search for instances of them in data sets or trying to blacklist malicious ones.

While there is some merit to this approach, it comes with all the known limitations of rules-and-signatures defenses, such as the overlaps in signatures, the inability to detect unknown threats, as well as the added complexity of having to maintain a database of known signatures.

JA3 in Darktrace

Darktrace creates JA3 hashes for every TLS/SSL connection it encounters. This is incredibly powerful in a number of ways. First, the JA3 can add invaluable context to a threat hunt. Second, Darktrace can also be queried to see if a particular JA3 was encountered in the network, thus providing actionable intelligence during incident response if JA3 IoCs are known to the incident responders.

Things become much more interesting once we apply our unsupervised machine learning to JA3: Darktrace’s AI algorithms autonomously detect which JA3s are anomalous for the network as a whole and which JA3s are unusual for specific devices.

It basically tells a cyber security expert: This JA3 (3e860202fc555b939e83e7a7ab518c38) has never been seen in the network before and it is only used by one device. It indicates that an application, which is used by nobody else on the network, is initiating TLS/SSL connections. In our experience, this is most often the case for malware or non-compliant software. At this stage, we are observing anomalous behavior.

Darktrace’s AI combines these IoCs (Unusual Network JA3, Unusual Device JA3, …) with many other weak indicators to detect the earliest signs of an emerging threat, including previously unknown threats, without using rules or hard-coded thresholds.

Catching Red-Teams and domain fronting with JA3

The following is an example where Darktrace detected a Red-Team’s C2 communication by observing anomalous JA3 behavior.

The unsupervised machine learning algorithms identified a desktop device using a JA3 that was 100% unusual for the network connecting to an external domain using a Let’s Encrypt certificate, which, along with self-signed certificates, is often abused by malicious actors. As well as the JA3, the domain was also 100% rare for the network – nobody else visited it:

It turned out that a Red-Team had registered a domain that was very similar to the victim’s legitimate domain: www.companyname[.]com (legitimate domain) vs. www.companyname[.]online (malicious domain). This was intentionally done to avoid suspicion and human analysis. Over a 7-day period in a 2,000-device environment, this was the only time that Darktrace flagged unusual behavior of this kind.

As the C2 traffic was encrypted (therefore no intrusion detection was possible on the payload) and the domain was non-suspicious (no reputation-based blacklisting worked), this C2 had remained undetected by the rest of the security stack.

Combining unsupervised machine learning with JA3 is incredibly powerful for the detection of domain fronting. Domain fronting is a popular technique to circumvent censorship and to hide C2 traffic. While some infrastructure providers take action to prevent domain fronting on their end, it is still prevalent and actively used by attackers.

The only agreed-upon method within wide parts of the cyber-security community to detect domain fronting appears to be TLS/SSL inspection. This usually involved breaking up encrypted communication to inspect the clear-text payloads. While this works, it commonly involves additional infrastructure, network restructuring and comes with privacy issues – especially in the context of GDPR.

Unsupervised machine learning makes the detection of domain fronting without having to break up encrypted traffic possible by combining unusual JA3 detection with other anomalies such as beaconing. A good start for a domain fronting threat hunt? A device beaconing to an anomalous CDN with an unusual JA3 hash.

Conclusion

JA3 is not a silver bullet to pre-empt malware compromise. As a signature-based solution, it shares the same limitations of all other defenses that rely on pre-identified threats or blacklists: having to play a constant game of catch-up with innovative attackers. However, as a novel means of identifying TLS/SSL applications, JA3 hashing can be leveraged as a powerful network behavioral indicator, an additional metric that can flag the use of unauthorized or risky software, or as a means of identifying emerging malware compromises in the initial stages of C2 communication. This is made possible through the power of unsupervised machine learning.

Max Heinemeyer

Max is a cyber security expert with over nine years’ experience in the field, specializing in network monitoring and offensive security. At Darktrace, Max works with strategic customers to help them investigate and respond to threats, as well as overseeing the cyber security analyst team in the Cambridge UK headquarters. Prior to his current role, Max led the Threat and Vulnerability Management department for Hewlett-Packard in Central Europe. In this role he worked as a white hat hacker, leading penetration tests and red team engagements. He was also part of the German Chaos Computer Club when he was still living in Germany. Max holds a MSc from the University of Duisburg-Essen and a BSc from the Cooperative State University Stuttgart in International Business Information Systems.

Ransomware one year after WannaCry: attack vectors still commonly exploited by attackers

Dave Palmer, Director of Technology | Wednesday May 9, 2018

Ransomware continues to be one of the most serious and disruptive cyber threats. The business models, motivations, and infection techniques of emerging campaigns have diversified, and new strands of ransomware continue to outpace the release of decryption tools. By 2019, global ransomware damage costs are expected to surpass $11.5 billion per year.

The three most memorable ransomware campaigns of 2017 - Wannacry, NotPetya, and Bad Rabbit - were ground-breaking in their scope, spread, and destructive power, demonstrating that every business, industry, and country is a potential victim. Although the damage caused by these attacks highlighted the importance of good cyber hygiene, many companies have struggled to address even the most widely reported vulnerabilities. As prevention is better than cure, this article will discuss some of the most common infection vectors and how the Darktrace Enterprise Immune System can assist security teams in catching ransomware threats.

Motivations: financial gain or wreaking havoc?

Ransomware is traditionally linked with making a quick buck by getting the victim to pay a set fee to unlock encrypted files. The phenomenon of ransomware-as-a-service has made this easier than ever before, as it has allowed virtually anyone to purchase ever more potent ransomware distribution kits on the Dark Web. The recent growth in cryptocurrencies has also made maintaining anonymity much easier than before, resulting in a definite increase in financially motivated cyber-criminals.

Regrettably, the goal of ransomware is no longer just to make money. NotPetya and other campaigns such as Ordinypt were designed to purposefully destroy data instead. Even though NotPetya provided its victims with payment instructions, it had no way of identifying who had actually made the payment. The uncertainty surrounding the recovery of lost files and the possibility of being associated with funding malicious organizations have therefore deterred many victims from meeting the ransom demands.

No matter how much a business tries to safeguard their assets, incidents are inevitable, and ransom attacks are an increasingly likely choice of criminal action. But it is now possible to identify in-progress attacks and handle them before they become a crisis.

Case Study 1: Executable file download from a compromised website

Many prolific ransomware strands have been distributed by phishing emails, infected file downloads, compromised websites, malvertising, and exploit kits. In many cases, ransomware is often downloaded and installed without the victim’s knowledge. To illustrate the ransomware download mechanics, we will analyze the life-cycle of a GandCrab incident. In the case study detailed below, the Darktrace Enterprise Immune System flagged a customer device retrieving an executable file from a previously unmonitored location following a redirection from another rare site.

The file containing ransomware was downloaded from a website registered to a Polish domain. Shortly after downloading the file, the customer’s device began reaching out to two locations which had not been contacted by any other network devices, nomoreransom.bit and bleepingcomputer.bit. Both are command and control servers for GandCrab ransomware. Once contacted, the malicious virus proceeded to encrypt files on the SMB server, adding the .GDCB (GandCrab) extension as it moved through the folders.

The virus modified the original file extensions in the encryption process.

Within seconds of the virus appearing on the company’s network, the Darktrace Cyber Analyst team alerted the security team of the threat. Preventative action was then taken, which allowed the threat to be contained within a timely manner.

Case Study 2: Bruteforcing Remote Desktop Protocol access

In addition to devising clever ways of downloading ransomware onto victim’s machines, some hackers have turned to bruteforcing Remote Desktop Protocol (RDP) access instead (HC7 & Lockcrypt). Exposing Remote Desktop services to the Internet is risky, as attackers can force access into a network by guessing login information and remotely exploiting a range of possible vulnerabilities and administrative tools in order to infect other available machines.

In another particularly serious breach, Darktrace detected a series of suspicious activities indicating that a malicious actor had taken control of a key server and was using it as a pivot point in order to move laterally throughout the network and install Remote Access Tools (RATs) on multiple devices.

In the initial stage of the attack, the Darktrace Enterprise Immune System observed over 400,000 incoming connections on a port that was targeting devices with RDP turned on and immediately flagged the first signs of a bruteforce attack.

Over 400,000 incoming RDP from multiple rare external devices.

The attack was successful; a compromised server was then used to retrieve malware that granted backdoor access and scanned the network for devices with open RDP channel. The hacker subsequently tunneled through the intermediary, gained control over multiple other machines, and installed third-party remote access software to all available devices.

Although most RDP bruteforcing incidents the Darktrace Enterprise Immune System observes do not escalate this far, the Darktrace Cyber Analyst team are constantly flagging instances of publicly accessible remote management services. To prevent ransomware that specifically exploits insecure RDP configuration, businesses should move these critical services to a virtual private network. Moreover, with Darktrace Antigena, Darktrace’s autonomous response solution, businesses can benefit from an added layer of protection. In this case, it would have blocked any anomalous RDP connections to the server, thus preventing any lateral movement throughout the network.

Dave Palmer

Dave is the Director of Technology at Darktrace, overseeing the mathematics and engineering teams and project strategies. With over 19 years of experience at the forefront of government intelligence operations, Dave has worked across UK intelligence agencies GCHQ and MI5, where he was responsible for delivering mission-critical infrastructure services, including replacing and securing entire global networks, the development of operational internet capabilities and the management of critical disaster recovery incidents. He acts as an advisor to cyber security start-ups and growth-stage companies from the UK Government’s Cyber Security Accelerator and CyLon. His insights on AI and the future of cyber security are also regularly featured in the UK media. He holds a first-class degree in Computer Science and Software Engineering from the University of Birmingham.