Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Written by
Max Heinemeyer
Global Field CISO
Written by
Adam Stevens
Director of Product, Cloud Security
Share
09
Aug 2023
TLDR: There are too many technical vulnerabilities and there is too little organizational context for IT teams to patch effectively. Attack path modelling provides the organizational context, allowing security teams to prioritize vulnerabilities. The result is a system where CVEs can be parsed in, organizational context added, and attack paths considered, ultimately providing a prioritized list of vulnerabilities that need to be patched.
Figure 1: The Darktrace user interface presents risk-prioritized vulnerabilities
This blog post explains how Darktrace addresses the challenge of vulnerability prioritization. Most of the industry focusses on understanding the technical impact of vulnerabilities globally (‘How could this CVE generally be exploited? Is it difficult to exploit? Are there pre-requisites to exploitation? …’), without taking local context of a vulnerability into account. We’ll discuss here how we create that local context through attack path modelling and map it to technical vulnerability information. The result is a stunningly powerful way to prioritize vulnerabilities.
We will explore:
1) The challenge and traditional approach to vulnerability prioritization 2) Creating local context through machine learning and attack path modelling 3) Examining the result – contextualized, vulnerability prioritization
The Challenge
Anyone dealing with Threat and Vulnerability Management (TVM) knows this situation:
You have a vulnerability scanning report with dozens or hundreds of pages. There is a long list of ‘critical’ vulnerabilities. How do you start prioritizing these vulnerabilities, assuming your goal is reducing the most risk?
Sometimes the challenge is even more specific – you might have 100 servers with the same critical vulnerability present (e.g. MoveIT). But which one should you patch first, as all of those have the same technical vulnerability priority (‘critical’)? Which one will achieve the biggest risk reduction (critical asset e.g.)? Which one will be almost meaningless to patch (asset with no business impact e.g.) and thus just a time-sink for the patch and IT team?
There have been recent improvements upon flat CVE-scoring for vulnerability prioritization by adding threat-intelligence about exploitability of vulnerabilities into the mix. This is great, examples of that additional information are Exploit Prediction Scoring System (EPSS) and Known Exploited Vulnerabilities Catalogue (KEV).
With CVE and CVSS scores we have the theoretical technical impact of vulnerabilities, and with EPSS and KEV we have information about the likelihood of exploitation of vulnerabilities. That’s a step forward, but still doesn’t give us any local context. Now we know even more about the global and generic technical risk of a vulnerability, but we still lack the local impact on the organization.
Let’s add that missing link via machine learning and attack path modelling.
Adding Attack Path Modelling for Local Context
To prioritize technical vulnerabilities, we need to know as much as we can about the asset on which the vulnerability is present in the context of the local organization. Is it a crown jewel? Is it a choke point? Does it sit on a critical attack path? Is it a dead end, never used and has no business relevance? Does it have organizational priority? Is the asset used by VIP users, as part of a core business or IT process? Does it share identities with elevated credentials? Is the human user on the device susceptible to social engineering?
Those are just a few typical questions when trying to establish local context of an asset. Knowing more about the threat landscape, exploitability, or technical information of a CVE won’t help answer any of the above questions. Gathering, evaluating, maintaining, and using this local context for vulnerability prioritization is the hard part. This local context often resides informally in the head of the TVM or IT team member, having been assembled by having been at the organization for a long time, ‘knowing’ systems, applications and identities in question and talking to asset and application owners if time permits. This does unfortunately not scale, is time-consuming and heavily dependent on individuals.
Understanding all attack paths for an organization provides this local context programmatically.
We discover those attack paths, and these are bespoke for each organization through Darktrace PREVENT™, using the following method (simplified):
1) Build an adaptive model of the local business. Collect, combine, and analyze (using machine learning and non-machine learning techniques) data from various data domains:
a. Network, Cloud, IT, and OT data (network-based attack paths, communication patterns, peer-groups, choke-points, …). Natively collected by Darktrace technology.
b. Email data (social engineering attack paths, phishing susceptibility, external exposure, security awareness level, …). Natively collected by Darktrace technology.
c. Identity data (account privileges, account groups, access levels, shared permissions, …). Collected via various integrations, e.g. Active Directory.
d. Attack surface data (internet-facing exposure, high-impact vulnerabilities, …). Natively collected by Darktrace technology.
e. SaaS information (further identity context). Natively collected by Darktrace
f. Vulnerability information (CVEs, CVSS, EPSS, KEV, …). Collected via integrations, e.g. Vulnerability Scanners or Endpoint products.
Figure 3: Darktrace PREVENT revealing each stage of an attack path
2) Understand what ‘crown jewels’ are and how to get to them. Calculate entity importance (user, technical asset), exposure levels, potential damage levels (blast radius) weakness levels, and other scores to identify most important entities and their relationships to each other (‘crown jewels’).
Various forms of machine learning and non-machine learning techniques are used to achieve this. Further details on some of the exact methods can be found here. The result is a holistic, adaptive and dynamic model of the organization that shows most important entities and how to get to them across various data domains.
The combination of local context and technical context, around the severity and likelihood of exploitation, creates the Darktrace Vulnerability Score. This enables effective risk-based prioritisation of CVE patching.
Figure 4: List of devices with the highest damage potential in the organization - local context
3) Map the attack path model of the organization to common cyber domain knowledge. We can then combine things like MITRE ATT&CK techniques with those identified connectivity patterns and attack paths – making it easy to understand which techniques, tools and procedures (TTPs) can be used to move through the organization, and how difficult it is to exploit each TTP.
Figure 5: An example attack path with associated MITRE techniques and difficulty scores for each TTP
We can now easily start prioritizing CVE patching based on actual, organizational risk and local context.
Bringing It All Together
Finally, we overlay the attack paths calculated by Darktrace with the CVEs collected from a vulnerability scanner or EDR. This can either happen as a native integration in Darktrace PREVENT, if we are already ingesting CVE data from another solution, or via CSV upload.
Figure 6: Darktrace's global CVE prioritization in action.
But you can also go further than just looking at the CVE that delivers the biggest risk reduction globally in your organization if it is patched. You can also look only at certain group of vulnerabilities, or a sub-set of devices to understand where to patch first in this reduced scope:
Figure 7: An example of the information Darktrace reveals around a CVE
This also provides the TVM team clear justification for the patch and infrastructure teams on why these vulnerabilities should be prioritized and what the positive impact will be on risk reduction.
Attack path modelling can be utilized for various other use cases, such as threat modelling and improving SOC efficiency. We’ll explore those in more depth at a later stage.
Want to explore more on using machine learning for vulnerability prioritization? Want to test it on your own data, for free? Arrange a demo today.
Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Customer Case Study: Leading Petrochemical Manufacturer
Headquartered in Saudi Arabia, this industry leading petrochemical manufacturer serves customers in more than 80 countries across diverse markets throughout Europe, Africa, Latin America, the Middle East, China, and Southeast Asia.
Cyber resiliency critical to growth strategy
This leading petrochemical manufacturer’s vision is to be one of the major global players in the production and marketing of designated petrochemicals and downstream products. The company aims to significantly increase its capacity to up to a million metric tons within the next few years.
With cyber-attacks on critical infrastructure increasing 30% globally last year, cyber resiliency is essential to supporting the company’s strategic business goals of:
Maximizing production through efficient asset utilization
Maximizing sales by conducting 90% of its business outside Saudi Arabia
Optimizing resources and processes by integrating with UN Global Compact principles for sustainability and efficiency
Growing its business portfolio by engaging in joint ventures to diversify production and add value to the economy
However, the industry leader faced several challenges in its drive to fortify its cybersecurity defenses.
Visibility gaps delay response time
The company’s existing security setup provided limited visibility to the in-house security team, hindering its ability to detect anomalous network and user activity in real time. This resulted in delayed responses to potential incidents, making proactive issue resolution difficult and any remediation in the event of a successful attack costly and time-consuming.
Manual detection drains resources
Without automated detection and response capabilities, the organization’s security team had to manually monitor for suspicious activity – a time-consuming and inefficient approach that strained resources and left the organization vulnerable. This made it difficult for the team to stay current with training or acquire new skills and certifications, which are core to the ethos of both the company’s owners and the team itself.
Cyber-attacks on critical infrastructure increasing
The petrochemical manufacturer is part of a broader ecosystem of companies, making the protection of its supply chain – both upstream and downstream – critical. With several manufacturing entities and multiple locations, the customer’s internal structure is complex and challenging to secure. As cyber-attacks on critical infrastructure escalate, it needed a more comprehensive approach to safeguard its business and the wider ecosystem.
Keeping and growing skills and focus in-house
To strengthen its cybersecurity strategy, the company considered two options:
Make a significant initial and ongoing investment in a Security Operations Center (SOC), which would involve skills development outside the company and substantial management overhead.
Use a combination of new, automated tools and an outsourced Managed Detection and Response (MDR) service to reduce the burden on internal security specialists and allow the company to invest in upskilling its staff so they can focus on more strategic tasks.
Faced with this choice between entirely outsourcing security and augmenting the security team with new capabilities, the customer chose the second option, selecting Darktrace to automate the company’s monitoring, detection, and response. Today, the petrochemical manufacturer is using:
To alleviate the burden on its lean security team, the company augmented its in-house capabilities with Darktrace’s Managed Detection & Response service. This support acts as an extension of its SOC, providing 24/7 monitoring, investigation, and escalation of high-priority threats. With Darktrace’s global SOC managing alert triage and autonomously containing threats, the organization’s internal team can focus on strategic initiatives. The result is a stronger security posture and increased capacity to proactively address evolving cyber risks – without expanding headcount or sacrificing visibility.
A unique approach to AI
In its search for a new security platform, the company’s Director of Information Technology said Darktrace’s autonomous response capability, coupled with Self-Learning AI-driven threat reduction, were two big reasons for selecting Darktrace over competing products and services.
AI was a huge factor – no one else was doing what Darktrace was doing with [AI].”
Demonstrated visibility
Before Darktrace, the customer had no visibility into the network activity to and from remote worker devices. Some employees need the ability to connect to its networks at any time and from any location, including the Director of Information Technology. The trial deployment of Darktrace / ENDPOINT was a success and gave the team peace of mind that, no matter the location or device, high-value remote workers were protected by Darktrace.
Modular architecture
Darktrace's modular architecture allowed the company to deploy security controls across its complex, multi-entity environment. The company’s different locations run on segregated networks but are still interconnected and need to be protected. Darktrace / NETWORK provides a unified view and coordinated security response across the organization’s entire network infrastructure, including endpoint devices.
Results
The petrochemical manufacturer is using Darktrace across all of its locations and has achieved total visibility across network and user activity. “Darktrace is increasing in value every day,” said the Director of Information Technology.
I don’t have a big team, and Darktrace makes our lives very, very easy, not least the automation of some of the tasks that require constant manual review.”
Time savings frees analysts to focus on proactive security
Darktrace / NETWORK provides continuous, AI-driven monitoring and analysis of the company’s network activity, user behavior, and threat patterns, establishing a baseline of what normal activity looks like, and then alerting analysts to any deviations from normal traffic, activity, and behaviors. Darktrace’s autonomous response capabilities speed up response to detected threats, meaning intervention from the security team is required for fewer incidents and alerts.
In October 2024 alone, Darktrace Cyber AI Analyst saved the team 810 investigation hours, and autonomously responded to 180 anomalous behaviors that were uncovered during the investigations. With Darktrace managing the majority of threat detection and response efforts, the security team has been able to change its day-to-day activity from manual review of traffic and alerts and belated response to activity, to proactively fortifying its detection and response posture and upskilling to meet evolving requirements.
Layered email protection reduces phishing threats
The company’s email infrastructure posed a challenge due to petrochemical industry regulations requiring on-premises email servers, with some security delivered via Microsoft Azure. By integrating Darktrace / EMAIL into the Azure stack, the organization has reduced the volume of phishing emails its users receive by 5%.
“Now we have one more layer of security related to email – every email goes through two filters. If something is not being caught or traced by Azure, it is being detected by Darktrace,” said the Director of Information Technology. “As a result, we’re now seeing only about 15% to 20% of the phishing emails we used to receive before implementing Darktrace.”
Preparing for a secure future
The time saved using Darktrace has helped the security team take proactive steps, including preparing for new cyber resilience regulations for Saudi Arabia’s Critical National Infrastructure, as mandated by the National Cybersecurity Authority (NCA).
“The team now has ample time to prepare policies and procedures that meet the new NCA regulations and, in some cases, enhance the requirements of the new law,” said the Director of Information Technology. “All of this is possible because they don’t need to keep watch; Darktrace takes on so much of that task for them.”
Tracking CVE-2025-31324: Darktrace’s detection of SAP Netweaver exploitation before and after disclosure
Introduction: Exploiting SAP platforms
Global enterprises depend extensively on SAP platforms, such as SAP NetWeaver and Visual Composer, to run critical business processes worldwide. These systems; however, are increasingly appealing targets for well-resourced adversaries:
In March 2025, CISA issued an alert confirming active exploitation of a 2017 SAP NetWeaver vulnerability (CVE‑2017‑12637), enabling attackers to perform directory traversal and exfiltrate sensitive files, including credentials, from internet-facing systems
CVE-2025-31324 affects SAP’s NetWeaver Visual Composer, a web-based software modeling tool. SAP NetWeaver is an application server and development platform that runs and connects SAP and non-SAP applications across different technologies [2]. It is commonly used by process specialists to develop application components without coding in government agencies, large enterprises, and by critical infrastructure operators [4].
CVE-2025-31324 affects SAP’s Netweaver Visual Composer Framework 7.1x (all SPS) and above [4]. The vulnerability in a Java Servlet (/irj/servlet_jsp) would enable an unauthorized actor to upload arbitrary files to the /developmentserver/metadatauploader endpoint, potentially resulting in remote code execution (RCE) and full system compromise [3]. The issue stems from an improper authentication and authorization check in the SAP NetWeaver Application Server Java systems [4].
What is the severity rating of CVE-2025-31324?
The vulnerability, first disclosed on April 24, 2025, carries the highest severity rating (CVSS v3 score: 10.0) and could allow remote attackers to upload malicious files without requiring authentication [1][5]. Although SAP released a workaround on April 8, many organizations are hesitant to take their business-critical SAP NetWeaver systems offline, leaving them exposed to potential exploitation [2].
How is CVE-2025-31324 exploited?
The vulnerability is exploitable by sending specifically crafted GET, POST, or HEAD HTTP requests to the /developmentserver/metadatauploader URL using either HTTP or HTTPS. Attackers have been seen uploading malicious files (.jsp, .java, or .class files to paths containing “\irj\servlet_jsp\irj\”), most of them being web shells, to publicly accessible SAP NetWeaver systems.
External researchers observed reconnaissance activity targeting this vulnerability in late January 2025, followed by a surge in exploitation attempts in February. The first confirmed compromise was reported in March [4].
Multiple threat actors have reportedly targeted the vulnerability, including Chinese Advanced Persistent Threats (APTs) groups Chaya_004 [7], UNC5221, UNC5174, and CL-STA-0048 [8], as well as ransomware groups like RansomEXX, also known as Storm-2460, BianLian [4] or Qilin [6] (the latter two share the same indicators of compromise (IoCs)).
Following the initial workaround published on April 8, SAP released a security update addressing CVE-2025-31324 and subsequently issued a patch on May 13 (Security Note 3604119) to resolve the root cause of the vulnerability [4].
Darktrace’s coverage of CVE-2025-31324 exploitation
Darktrace has observed activity indicative of threat actors exploiting CVE-2025-31324, including one instance detected before the vulnerability was publicly disclosed.
In April 2025, the Darktrace Threat Research team investigated activity related to the CVE-2025-31324 on SAP devices and identified two cases suggesting active exploitation of the vulnerability. One case was detected prior to the public disclosure of the vulnerability, and the other just two days after it was published.
Early detection of CVE 2025-31324 by Darktrace
Figure 1: Timeline of events for an internet-facing system, believed to be a SAP device, exhibiting activity indicative of CVE-2025-31324 exploitation.
On April 18, six days prior to the public disclosure of CVE-2025-31324, Darktrace began to detect unusual activity on a device belonging to a logistics organization in the Europe, the Middle East and Africa (EMEA) region. Multiple IoCs observed during this incident have since been linked via OSINT to the exploitation of CVE-2025-31324. Notably, however, this reporting was not available at the time of detection, highlighting Darktrace’s ability to detect threats agnostically, without relying on threat intelligence.
The device was observed making domain name resolution request for the Out-of-Band Application Security Testing (OAST) domain cvvr9gl9namk9u955tsgaxy3upyezhnm6.oast[.]online. OAST is often used by security teams to test if exploitable vulnerabilities exist in a web application but can similarly be used by threat actors for the same purpose [9].
Four days later, on April 22, Darktrace observed the same device, an internet-facing system believed to be a SAP device, downloading multiple executable (.exe) files from several Amazon Simple Storage Service (S3). Darktrace’s Threat Research team later found these files to be associated with the KrustyLoader malware [23][24][25].
KrustyLoader is known to be associated with the Chinese threat actor UNC5221, also known as UTA0178, which has been reported to aggressively target devices exposed to the internet [10] [14] [15]. It is an initial-stage malware which downloads and launches a second-stage payload – Sliver C2. Sliver is a similar tool to Cobalt Strike (an open-source post-exploitation toolkit). It is used for command-and-control (C2) connections [11][12]13]. After its successful download, KrustyLoader deletes itself to evade detection. It has been reported that multiple Chinese APT groups have deployed KrustyLoader on SAP Netweaver systems post-compromise [8].
The actors behind KrustyLoader have also been associated with the exploitation of zero-day vulnerabilities in other enterprise systems, including Ivanti devices [12]. Notably, in this case, one of the Amazon S3 domains observed (abode-dashboard-media.s3.ap-south-1.amazonaws[.]com ) had previously been investigated by Darktrace’s Threat Research team as part of their investigation into Ivanti Connect Secure (CS) and Policy Secure (PS) appliances.
In addition to the download of known malicious files, Darktrace also detected new IoCs, including several executable files that could not be attributed to any known malware families or previous attacks, and for which no corresponding OSINT reporting was available.
Post-CVE publication detection
Exploit Validation
Between April 27 and 29, Darktrace observed unusual activity from an SAP device on the network of a manufacturing customer in EMEA.
Figure 2: Darktrace / NETWORK’s detection of an SAP device performing a large volume of suspicious activity between April 27 and April 29.
The device was observed making DNS requests for OAST domains (e.g. aaaaaaaa.d06qqn7pu5a6u25tv9q08p5xhbjzw33ge.oast[.]online and aaaaaaaaaaa.d07j2htekalm3139uk2gowmxuhapkijtp.oast[.]pro), suggesting that a threat actor was testing for exploit validation [9].
Figure 3: Darktrace / NETWORK’s detection of a SAP device making suspicious domain name resolution requests for multiple OAST domains.
Privilege escalation tool download attempt
One day later, Darktrace observed the same device attempting to download an executable file from hxxp://23.95.123[.]5:666/xmrigCCall/s.exe (SHA-1 file hash: e007edd4688c5f94a714fee036590a11684d6a3a).
Darktrace / NETWORK identified the user agents Microsoft-CryptoAPI/10.0 and CertUtil URL Agent during the connections to 23.95.123[.]5. The connections were made over port 666, which is not typically used for HTTP connections.
Multiple open-source intelligence (OSINT) vendors have identified the executable file as either JuicyPotato or SweetPotato, both Windows privilege escalation tools[16][17][18][19]. The file hash and the unusual external endpoint have been associated with the Chinese APT group Gelsemium in the past, however, many threat actors are known to leverage this tool in their attacks [20] [21].
Figure 4: Darktrace’s Cyber AI Analyst’s detection of a SAP device downloading a suspicious executable file from hxxp://23.95.123[.]5:666/xmrigCCall/s.exe on April 28, 2025.
Darktrace deemed this activity highly suspicious and triggered an Enhanced Monitoring model alert, a high-priority security model designed to detect activity likely indicative of compromise. As the customer was subscribed to the Managed Threat Detection service, Darktrace’s Security Operations Centre (SOC) promptly investigated the alert and notified the customer for swift remediation. Additionally, Darktrace’s Autonomous Response capability automatically blocked connections to the suspicious IP, 23.95.123[.]5, effectively containing the compromise in its early stages.
Figure 5: Actions taken by Darktrace’s Autonomous Response to block connections to the suspicious external endpoint 23.95.123[.]5. This event log shows that the connections to 23.95.123[.]5 were made over a rare destination port for the HTTP protocol and that new user agents were used during the connections.
Conclusion
The exploitation of CVE-2025-31324 to compromise SAP NetWeaver systems highlights the persistent threat posed by vulnerabilities in public-facing assets. In this case, threat actors leveraged the flaw to gain an initial foothold, followed by attempts to deploy malware linked to groups affiliated with China [8][20].
Crucially, Darktrace demonstrated its ability to detect and respond to emerging threats even before they are publicly disclosed. Six days prior to the public disclosure of CVE-2025-31324, Darktrace detected unusual activity on a device believed to be a SAP system, which ultimately represented an early detection of the CVE. This detection was made possible through Darktrace’s behavioral analysis and anomaly detection, allowing it to recognize unexpected deviations in device behavior without relying on signatures, rules or known IoCs. Combined with its Autonomous Response capability, this allowed for immediate containment of suspicious activity, giving security teams valuable time to investigate and mitigate the threat.
Credit to Signe Zaharka (Principal Cyber Analyst), Emily Megan Lim, (Senior Cyber Analyst) and Ryan Traill (Analyst Content Lead)
Appendices
List of IoCs
23.95.123[.]5:666/xmrigCCall/s.exe - URL- JuicyPotato/SweetPotato - high confidence
29274ca90e6dcf5ae4762739fcbadf01- MD5 file hash - JuicyPotato/SweetPotato - high confidence
![Actions taken by Darktrace’s Autonomous Response to block connections to the suspicious external endpoint 23.95.123[.]5. This event log shows that the connections to 23.95.123[.]5 were made over a rare destination port for the HTTP protocol and that new user agents were used during the connections.](https://cdn.prod.website-files.com/626ff4d25aca2edf4325ff97/685195a1592b0029efa4fd53_Screenshot%202025-06-17%20at%209.19.36%E2%80%AFAM.png)