Blog
/
Network
/
April 27, 2022

How Darktrace AI Blocked Emotet Malspam

Explore Darktrace AI's success in combating Emotet malspam, enhancing security and minimizing risks with cutting-edge artificial intelligence.
Inside the SOC
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
Zoe Tilsiter
Cyber Analyst
Default blog image
27
Apr 2022


In January 2021, it was lauded that an international collaborative law enforcement operation had successfully dismantled Emotet’s infrastructure. This was one of the most prolific malware and banking Trojans which led to sensitive data loss, significant financial loss and reputational damage for its victims since early deployment in 2014.1

However, since November 2021, there have been signs of Emotet’s resurgence. Emotet has supposedly leveraged its former partner operators such as Trickbot, also discussed in another Darktrace blog, to rebuild its infrastructure by using already infected machines to download the new Emotet binary.2

Early signs of Emotet’s return appear to be synonymous with its original kill chain and attack vectors. Malware is deployed, compromising a device as a zombie machine. This device is then used to send outbound malspam campaigns. These campaigns can be masked as application installer packages or fake reply email chains to give the spam credibility. Once the malware spreads through this spam, it then attempts to infect other devices – both internally and outbound in other networks.3

In February 2022, Darktrace detected elements of this kill chain in a customer’s environment, notably observing the large volume of SMTP connections which are characteristic of an outbound spam campaign.

Figure 1: Timeline of attack showing the Emotet intrusion progress along the kill chain
Figure 2: A screenshot from VirusTotal, showing that the rare endpoint has been flagged as malicious by other security vendors


Bypassing the rest of the security stack

The attack used Living-off-the-Land techniques by making PowerShell connections via pre-existing user agents within the network. As PowerShell connections can be used for legitimate reasons, this activity appeared to bypass the rest of the customer’s security stack and was likely seen as approved by their tools. However, Darktrace detected that the device was using the PowerShell user agent to connect to an external location. This is rare in comparison to wider network behavior.

The customer’s pre-existing security did not block the outgoing SMTP connections made by the compromised device on unusual ports. However, Darktrace Antigena blocked 71% of outgoing connections on mail ports 25 and 587, significantly reducing the scale of the spam dissemination.

Darktrace insights and services

Darktrace quickly detected a range of anomalous behaviors from the new PowerShell use, uptake in C2 beaconing activity and spam. This can be highlighted via the spike in model breaches (Figure 3). Darktrace’s Cyber AI Analyst also launched an investigation into the device’s suspicious network scanning activity. This was essential for generating an incident summary which outlined the investigation process and technical details needed for the organization’s security team to act quickly (Figure 4).

Throughout the incident, Antigena autonomously responded to the initial breach device to enforce its ‘pattern of life’ without interrupting business processes. This significantly reduced the scope of the compromise by halting further lateral movement. In response to the malicious outbound email spam, Antigena enforced the device’s usual ‘pattern of life’ for thirty minutes and blocked connections to ports 25, 80 and 587 for one hour (Figure 5). Against the command-and-control activity, connections to 91.207.181[.]106 via port 8080 were also blocked for three hours.

The customer’s subscription to Darktrace’s Proactive Threat Notification (PTN) and Ask the Expert (ATE) services meant that this compromise was assisted by additional triage and alerting. PTN ensured that the Darktrace SOC team were quickly alerted to the breach, enabling analysts to perform a detailed investigation alongside the customer’s own security team. Simultaneously, the ATE service ensured the customer was provided with additional information to ensure the threat was less likely to happen again. This equipped the team with the vital information needed for them to act, and to restore quickly and precisely.

Figure 3: Darktrace reveals an anomalous spike in the device’s activity and associated model breaches during the attack period, represented by the dots on the graph


Figure 4: Excerpt of the AI Analyst report of the breach device’s network scanning activity
Figure 5: Antigena Network blocking external connection activity and enforcing the device’s ‘pattern of life’


The resurgence of Emotet shows how email continues to act as a crucial attack vector and source of compromise. In particular, widespread malspam campaigns remain adaptable and effective. The incident in this blog is yet another example highlighting the alarming mutability and networked nature of malware organizations. This allows them to return, even long after their dismantling. Fortunately, in this incident, Autonomous Response enabled this Emotet compromise to be minimized, while PTN and ATE services alerted and further supported the security team throughout.

Appendix

Darktrace model breaches

·    Device / Multiple Lateral Movement Model Breaches

·    Device / Large Number of Model Breaches

·    Device / Suspicious Network Scan Activity

·    Device / Network Scan

·    Device / External Address Scan

·    Device / Multiple C2 Model Breaches

·    Device / Large Number of Connections to New Endpoints

·    Device / Increased External Connectivity

·    Device / New User Agent and New IP

·    Device / New PowerShell User Agent

·    Compromise / Suspicious Beaconing Behavior

·    Compromise / Beacon to Young Endpoint

·    Compromise / Agent Beacon to New Endpoint

·    Compromise / Sustained SSL or HTTP Increase

·    Compromise / Suspicious Spam Activity

·    Anomalous Connection / Possible Outbound Spam

·    Anomalous Connection / Suspicious Expired SSL

·    Anomalous Connection / Rare External SSL Self-Signed

·    Anomalous Connection / Suspicious Self-Signed SSL

·    Anomalous Connection / Anomalous SSL without SNI to New External

·    Anomalous Connection / PowerShell to Rare External

·    AI Analyst / AI Analyst Investigation

·    Unusual Activity / Unusual External Activity

IoCs

MITRE ATT&CK Techniques Observed

Footnotes

1. https://www.cisa.gov/uscert/ncas/alerts/TA18-201A

2. https://blog.malwarebytes.com/threat-intelligence/2021/11/trickbot-helps-emotet-come-back-from-the-dead/

3. https://www.kaspersky.com/resource-center/threats/emotet

Inside the SOC
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
Zoe Tilsiter
Cyber Analyst

More in this series

No items found.

Blog

/

/

June 24, 2026

A New Security Challenge: The Curious Case of Prompt Language Analysis

Default blog imageDefault blog image

Why prompt analysis is emerging as a key AI security challenge

If securing AI has been one of the defining cybersecurity conversations of the past year, prompt analysis is quickly becoming one of its most interesting frontiers.

Security leaders are under pressure to understand how AI is being used across the business. In some organizations, that means governing employee use of chatbots. In others, it means overseeing copilots embedded into SaaS platforms, monitoring coding assistants, or assessing the growing footprint of autonomous agents. However different these use cases may appear on the surface, they share a common factor: humans and machines are usually interacting with enterprise systems through language.  

How prompt language differs from traditional security telemetry

For years, defenders have become used to working with familiar forms of telemetry: email traffic, network connections, API calls, endpoint processes, authentication events. Prompt language is different. It is not simply another log source. It is an expression of intent, instruction, curiosity, urgency, and sometimes manipulation. It reflects the end-goal of a user or agent, but not always with enough surrounding context to interpret the risk correctly.

Why existing security approaches only partially explain prompt risk

A growing number of vendors are approaching the task of securing AI from the angle they know best. Perimeter vendors are extending web or browser controls into AI usage. Identity vendors are emphasizing agent permissions and access governance. Data security and DLP providers are focusing on content inspection and exfiltration risk. All of these perspectives matter, but individually can’t fully explain the problem.

The challenge with securing AI is not just that a new application category has emerged. It is that language has become a new operating layer in the enterprise.

Employees now use prompts to summarize documents, generate code, analyze spreadsheets, query internal knowledge, and trigger multi-step actions through agents. In each case, prompt language acts as the interface between human intent and machine execution. That makes prompts incredibly valuable from a security perspective as they can hint at misuse, policy violations, data exposure, or attempts to circumvent controls. However, they can also be deeply ambiguous when viewed in isolation. That ambiguity is the heart of the issue.

Prompts as behavioral signals, not just text to classify

A prompt by itself tells you what was asked. It does not necessarily tell you whether the request is expected, risky, accidental, or entirely legitimate in context. Two nearly identical prompts can carry very different meanings depending on the role and function of who issued them, what systems they can access, and what actions followed. In other words, prompts are not just text to classify. They are behavioral signals to interpret.

Example: How context changes prompt risk entirely

Consider a common enterprise scenario. An employee is pulled into a new project with an aggressive deadline. Almost overnight, their use of AI tools spikes. They begin prompting more frequently, working across unfamiliar documents, querying new data sources, and interacting with more systems than usual to accelerate delivery. Viewed narrowly, this may look suspicious. Prompt volume increases, file access patterns change, API and SaaS activity rise. From some vantage points, it may resemble insider risk or unmanaged AI usage.

But now add context. Imagine that, earlier that day, the employee received instructions from a senior leader asking them to support a time-sensitive initiative. Their communication history shows that this leader is a legitimate reporting-line superior. Their recent collaboration patterns align with the new project team. Their subsequent activity, while unusual for that individual’s baseline, is consistent with the business task they were assigned.

What initially looked like a risk event may actually be a normal response to business pressure. Without the surrounding context of communication, organizational relationships, and broader behavioral patterns, prompt activity alone could generate more noise than insight.

The reverse is also true. A prompt may appear benign on the surface while the context around it suggests elevated risk. A request that seems routine could originate from a compromised user, a newly connected external agent, a shadow AI workflow, or a user acting outside their normal role. The language itself may not contain anything obviously malicious, but the surrounding conditions may tell a very different story.

What security teams need to analyze prompts effectively

The future of prompt analysis is not just about understanding language. It is about understanding language in context.

To do that well, security teams need more than prompt inspection. They need to understand:

  • Who is issuing the prompt, whether human or agent
  • How that identity normally behaves across the enterprise
  • What systems, data, and workflows are connected to the interaction
  • Which relationships and communications explain the surrounding activity
  • Whether the downstream actions align with expected business behavior

When those layers are absent, prompt analysis can become another isolated control surface: useful in theory, but limited in practice. Security teams may detect unusual wording but miss the operational function behind it, overreact to benign changes in behavior, or miss subtle misuse because the prompt itself did not appear dangerous.

How organizations should think about prompt analysis going forward

Security teams have seen this pattern before. In the cloud, posture without runtime context left important gaps. In identity, access control without behavioral understanding missed misuse that looked legitimate on paper. In data security, content inspection without business context often created friction without resolving risk. AI is exposing the same lesson again: controls are strongest when they are coordinated, not isolated. As organizations work to secure AI and identify gaps across their security operations, prompt analysis will become an increasingly important source of insight, but only as part of a broader strategy.

Prompt analysis will undoubtedly become more common, as prompts are one of the clearest windows into how people and agents are using AI systems. However, what matters most is not simply collecting prompts or filtering dangerous phrases, but being able to place that language inside a wider behavioral and operational picture.

Organizations that already have a broader understanding of how work gets done across the enterprise will be better positioned to make sense of prompt language as this category matures. They will be better able to distinguish urgency from abuse, experimentation from exfiltration, and productive AI adoption from hidden risk.

Figure 1: Darktrace / SECURE AI reconstructs the full sequence of events, showing every user and agent interaction in context, with risky prompts highlighted and categorized, including PII, sensitive data, and other policy violations.

At Darktrace, this is the key lesson emerging from the market: prompt language does matter, but it does not stand alone. It is most valuable when treated as a new behavioral input that can enrich understanding across the enterprise, not as a self-contained source of truth.

Why prompts become less useful when analyzed in isolation

The curious case of prompt language analysis, then, is this: the more important prompts become, the less useful they are in a vacuum.

The real opportunity is not just to see what was asked. It is to understand why it was asked, what it meant in that moment, and what happened next.

For a deeper look at how organizations are approaching this challenge from the strengths of prompt analysis to its limitations in isolation see Prompt Security in Enterprise AI: Strengths, Weaknesses, and Common Approaches, which expands on the role prompt-level controls play within a broader, context-driven security strategy.

Continue reading
About the author
Nabil Zoldjalali
VP, Field CISO

Blog

/

/

June 23, 2026

Advancing the Use of Frontier AI in Cybersecurity: Darktrace Joins the OpenAI Daybreak Cyber Partner Program to Explore Defensive AI Integrations

Default blog imageDefault blog image

Darktrace joins the OpenAI Daybreak Cyber Partner Program

Today, we announced that Darktrace is joining the OpenAI Daybreak Cyber Partner Program. We’ll be partnering with OpenAI to explore how their cyber capabilities can be integrated within Darktrace products and services to bring new capabilities to our customers.

This partnership is an exciting opportunity to bring together Darktrace’s behavioral AI modelling of the organization with OpenAI’s advanced contextual capabilities to create a new level of understanding for security teams. To understand the impact, it’s helpful to start with how we think about the problem.  

At Darktrace, we built our AI in support of the core belief that cybersecurity needs to understand the business it is defending. That's why our Self-Learning AI is designed to help organizations understand normal and abnormal behavior for each organization across their digital environment, including users and identities, networks and cloud, email and collaboration tools, and now AI systems and agents with the rollout of Darktrace / SECURE AI™.  

Our goal was never simply to spot known attacks faster. It was to help defenders understand how their organization behaves, potential risks and impact, and where disruption could take hold so they could prepare for the unknown threats that they may not have seen or even imagined before.  

That’s exactly what is happening across the threat landscape today. Attacks keep changing; techniques shift, infrastructure evolves, and attackers move with more speed, precision, and context. And now they have even more AI and automation on their side. Attackers are exploiting identities, trusted services, SaaS applications, and business workflows. They are not always breaking in; often, the threat may come from within the organization in the form of insider threat or even rogue agents.  

In this reality, defenders need a combination of deep AI modelling of the organization and AI that can connect identified threats to concrete business context, translating this information into real world value, and allow action before risk becomes disruption.

That is the opportunity we see in partnering with OpenAI.  

What is the OpenAI Daybreak Cyber Partner Program and why is Darktrace joining

The OpenAI Daybreak Cyber Partner Program is focused on advancing the safe use of AI for cybersecurity. As part of the program’s next phase, OpenAI is working with a select group of trusted partners including Darktrace on scoped product integrations, managed services, and partner-delivered defensive capabilities. We’ll be exploring how OpenAI’s advanced frontier AI capabilities can support defenders in the tools and workflows they already use each day.

For Darktrace, this is a natural extension of our expertise and the work we have been doing for a decade: safely and securely applying the most effective AI techniques in combination to understand organizations, detecting malicious activity at the earliest indicators, and helping cyber defenders act faster.  

By using the advanced models and more precise safeguards available in the OpenAI Daybreak Cyber Partner Program, Darktrace and OpenAI will combine Darktrace’s real-time behavioral understanding of an organization's digital estate with OpenAI's ability to interpret wider business context.  

This is a unique and powerful combination of insights that could give organizations deeper context on technical risk and help them prioritize workloads and investigations based on potential impact to revenue, operations, and resilience. It can also provide security teams and executives with intelligence into which events matter most to the business, why they matter, and what action to take. Not just finding, for instance, that an agent is compromised, but highlighting that the compromised agent could shut down order fulfilment within the next three hours.  

Why the Darktrace and OpenAI partnership matters for defenders

Security teams today have more attack surface, more complex environments to protect, and an increasing volume of threats. The ability to act quickly is critical, but they also need to be able to focus on the risks that could have the greatest business impact.

That is especially important as attackers use AI to scale phishing, automate reconnaissance, find weaknesses, and blend into normal business activity. At the same time, organizations and their employees are using AI to innovate, which introduces an even broader attack surface and new set of risks. Defenders need AI that can operate across the same complexity, but safely, transparently, and in service of building more resilience. And they need a way to safely adopt, govern, and defend AI across their organizations.

Joining the OpenAI Daybreak Cyber Partner Program is another step in that direction. We are still early in this work, and we will take a careful, disciplined approach. But the direction is clear: protecting organizations requires AI that understands the business, not just the attack.

At Darktrace, that is exactly where we remain focused and why we are so excited about this partnership with OpenAI.  

[related-resource]

Continue reading
About the author
Your data. Our AI.
Elevate your network security with Darktrace AI