September 3, 2025

From PowerShell to Payload: Darktrace’s Detection of a Novel Cryptomining Malware

Cryptojacking attacks are rising as threat actors exploit hard-to-detect cryptomining malware. Learn how Darktrace detected and contained a cryptojacking attempt in its early stages using Autonomous Response, with expert analysis of the malware itself revealing insights into a novel cryptomining strain.

No items found.

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.

No items found.

Sep 2025

What is Cryptojacking?

Cryptojacking remains one of the most persistent cyber threats in the digital age, showing no signs of slowing down. It involves the unauthorized use of a computer or device’s processing power to mine cryptocurrencies, often without the owner’s consent or knowledge, using cryptojacking scripts or cryptocurrency mining (cryptomining) malware [1].

Unlike other widespread attacks such as ransomware, which disrupt operations and block access to data, cryptomining malware steals and drains computing and energy resources for mining to reduce attacker’s personal costs and increase “profits” earned from mining [1]. The impact on targeted organizations can be significant, ranging from data privacy concerns and reduced productivity to higher energy bills.

As cryptocurrency continues to grow in popularity, as seen with the ongoing high valuation of the global cryptocurrency market capitalization (almost USD 4 trillion at time of writing), threat actors will continue to view cryptomining as a profitable venture [2]. As a result, illicit cryptominers are being used to steal processing power via supply chain attacks or browser injections, as seen in a recent cryptojacking campaign using JavaScript [3][4].

Therefore, security teams should maintain awareness of this ongoing threat, as what is often dismissed as a "compliance issue" can escalate into more severe compromises and lead to prolonged exposure of critical resources.

While having a security team capable of detecting and analyzing hijacking attempts is essential, emerging threats in today’s landscape often demand more than manual intervention.

This blog will discuss Darktrace’s successful detection of the malicious activity, the role of Autonomous Response in halting the cryptojacking attack, include novel insights from Darktrace’s threat researchers on the cryptominer payload, showing how the attack chain was initiated through the execution of a PowerShell-based payload.

Darktrace’s Coverage of Cryptojacking via PowerShell

In July 2025, Darktrace detected and contained an attempted cryptojacking incident on the network of a customer in the retail and e-commerce industry.

The threat was detected when a threat actor attempted to use a PowerShell script to download and run NBMiner directly in memory.

The initial compromise was detected on July 22, when Darktrace / NETWORK observed the use of a new PowerShell user agent during a connection to an external endpoint, indicating an attempt at remote code execution.

Specifically, the targeted desktop device established a connection to the rare endpoint, 45.141.87[.]195, over destination port 8000 using HTTP as the application-layer protocol. Within this connection, Darktrace observed the presence of a PowerShell script in the URI, specifically ‘/infect.ps1’.

Darktrace’s analysis of this endpoint (45.141.87[.]195[:]8000/infect.ps1) and the payload it downloaded indicated it was a dropper used to deliver an obfuscated AutoIt loader. This attribution was further supported by open-source intelligence (OSINT) reporting [5]. The loader likely then injected NBMiner into a legitimate process on the customer’s environment – the first documented case of NBMiner being dropped in this way.

Figure 1: Darktrace’s detection of a device making an HTTP connection with new PowerShell user agent, indicating PowerShell abuse for command-and-control (C2) communications.

Script files are often used by malicious actors for malware distribution. In cryptojacking attacks specifically, scripts are used to download and install cryptomining software, which then attempts to connect to cryptomining pools to begin mining operations [6].

Inside the payload: Technical analysis of the malicious script and cryptomining loader

To confidently establish that the malicious script file dropped an AutoIt loader used to deliver the NBMiner cryptominer, Darktrace’s threat researchers reverse engineered the payload. Analysis of the file ‘infect.ps1’ revealed further insights, ultimately linking it to the execution of a cryptominer loader.

Figure 2: Screenshot of the ‘infect.ps1’ PowerShell script observed in the attack.

The ‘infect.ps1’ script is a heavily obfuscated PowerShell script that contains multiple variables of Base64 and XOR encoded data. The first data blob is XOR’d with a value of 97, after decoding, the data is a binary and stored in APPDATA/local/knzbsrgw.exe. The binary is AutoIT.exe, the legitimate executable of the AutoIt programming language. The script also performs a check for the existence of the registry key HKCU:\\Software\LordNet.

The second data blob ($cylcejlrqbgejqryxpck) is written to APPDATA\rauuq, where it will later be read and XOR decoded. The third data blob ($tlswqbblxmmr)decodes to an obfuscated AutoIt script, which is written to %LOCALAPPDATA%\qmsxehehhnnwioojlyegmdssiswak. To ensure persistence, a shortcut file named xxyntxsmitwgruxuwqzypomkhxhml.lnk is created to run at startup.

Figure 3: Screenshot of second stage AutoIt script.

The observed AutoIt script is a process injection loader. It reads an encrypted binary from /rauuq in APPDATA, then XOR-decodes every byte with the key 47 to reconstruct the payload in memory. Next, it silently launches the legitimate Windows app ‘charmap.exe’ (Character Map) and obtains a handle with full access. It allocates executable and writable memory inside that process, writes the decrypted payload into the allocated region, and starts a new thread at that address. Finally, it closes the thread and process handles.

The binary that is injected into charmap.exe is 64-bit Windows binary. On launch, it takes a snapshot of running processes and specifically checks whether Task Manager is open. If Task Manager is detected, the binary kills sigverif.exe; otherwise, it proceeds. Once the condition is met, NBMiner is retrieved from a Chimera URL (https://api[.]chimera-hosting[.]zip/frfnhis/zdpaGgLMav/nbminer[.]exe) and establishes persistence, ensuring that the process automatically restarts if terminated. When mining begins, it spawns a process with the arguments ‘-a kawpow -o asia.ravenminer.com:3838 -u R9KVhfjiqSuSVcpYw5G8VDayPkjSipbiMb.worker -i 60’ and hides the process window to evade detection.

The program includes several evasion measures. It performs anti-sandboxing by sleeping to delay analysis and terminates sigverif.exe (File Signature Verification). It checks for installed antivirus products and continues only when Windows Defender is the sole protection. It also verifies whether the current user has administrative rights. If not, it attempts a User Account Control (UAC) bypass via Fodhelper to silently elevate and execute its payload without prompting the user. The binary creates a folder under %APPDATA%, drops rtworkq.dll extracted from its own embedded data, and copies ‘mfpmp.exe’ from System32 into that directory to side-load ‘rtworkq.dll’. It also looks for the registry key HKCU\Software\kap, creating it if it does not exist, and reads or sets a registry value it expects there.

Zooming Out: Darktrace Coverage of NBMiner

Darktrace’s analysis of the malicious PowerShell script provides clear evidence that the payload downloaded and executed the NBMiner cryptominer. Once executed, the infected device is expected to attempt connections to cryptomining endpoints (mining pools). Darktrace initially observed this on the targeted device once it started making DNS requests for a cryptominer endpoint, “gulf[.]moneroocean[.]stream” [7], one minute after the connection involving the malicious script.

Figure 5: Darktrace Advanced Search logs showcasing the affected device making a DNS request for a Monero mining endpoint.

Though DNS requests do not necessarily mean the device connected to a cryptominer-associated endpoint, Darktrace detected connections to the endpoint specified in the DNS Answer field: monerooceans[.]stream, 152.53.121[.]6. The attempted connections to this endpoint over port 10001 triggered several high-fidelity model alerts in Darktrace related to possible cryptomining mining activity. The IP address and destination port combination (152.53.121[.]6:10001) has also been linked to cryptomining activity by several OSINT security vendors [8][9].

Figure 6: Darktrace’s detection of a device establishing connections with the Monero Mining-associated endpoint, monerooceans[.]stream over port 10001.

Darktrace / NETWORK grouped together the observed indicators of compromise (IoCs) on the targeted device and triggered an additional Enhanced Monitoring model designed to identify activity indicative of the early stages of an attack. These high-fidelity models are continuously monitored and triaged by Darktrace’s SOC team as part of the Managed Threat Detection service, ensuring that subscribed customers are promptly notified of malicious activity as soon as it emerges.

Figure 7: Darktrace’s correlation of the initial PowerShell-related activity with the cryptomining endpoint, showcasing a pattern indicative of an initial attack chain.

Darktrace’s Cyber AI Analyst launched an autonomous investigation into the ongoing activity and was able to link the individual events of the attack, encompassing the initial connections involving the PowerShell script to the ultimate connections to the cryptomining endpoint, likely representing cryptomining activity. Rather than viewing these seemingly separate events in isolation, Cyber AI Analyst was able to see the bigger picture, providing comprehensive visibility over the attack.

Figure 8: Darktrace’s Cyber AI Analyst view illustrating the extent of the cryptojacking attack mapped against the Cyber Kill Chain.

Darktrace’s Autonomous Response

Fortunately, as this customer had Darktrace configured in Autonomous Response mode, Darktrace was able to take immediate action by preventing the device from making outbound connections and blocking specific connections to suspicious endpoints, thereby containing the attack.

Specifically, these Autonomous Response actions prevented the outgoing communication within seconds of the device attempting to connect to the rare endpoints.

Figure 10: Darktrace’s Autonomous Response blocked connections to the mining-related endpoint within a second of the initial connection.

Additionally, the Darktrace SOC team was able to validate the effectiveness of the Autonomous Response actions by analyzing connections to 152.53.121[.]6 using the Advanced Search feature. Across more than 130 connection attempts, Darktrace’s SOC confirmed that all were aborted, meaning no connections were successfully established.

Figure 11: Advanced Search logs showing all attempted connections that were successfully prevented by Darktrace’s Autonomous Response capability.

Conclusion

Cryptojacking attacks will remain prevalent, as threat actors can scale their attacks to infect multiple devices and networks. What’s more, cryptomining incidents can often be difficult to detect and are even overlooked as low-severity compliance events, potentially leading to data privacy issues and significant energy bills caused by misused processing power.

Darktrace’s anomaly-based approach to threat detection identifies early indicators of targeted attacks without relying on prior knowledge or IoCs. By continuously learning each device’s unique pattern of life, Darktrace can detect subtle deviations that may signal a compromise.

In this case, the cryptojacking attack was quickly identified and mitigated during the early stages of malware and cryptomining activity. Darktrace's Autonomous Response was able to swiftly contain the threat before it could advance further along the attack lifecycle, minimizing disruption and preventing the attack from potentially escalating into a more severe compromise.

Credit to Keanna Grelicha (Cyber Analyst) and Tara Gould (Threat Research Lead)

Appendices

Darktrace Model Detections

NETWORK Models:

· Compromise / High Priority Crypto Currency Mining (Enhanced Monitoring Model)

· Device / Initial Attack Chain Activity (Enhanced Monitoring Model)

· Compromise / Suspicious HTTP and Anomalous Activity (Enhanced Monitoring Model)

· Compromise / Monero Mining

· Anomalous File / Script from Rare External Location

· Device / New PowerShell User Agent

· Anomalous Connection / New User Agent to IP Without Hostname

· Anomalous Connection / Powershell to Rare External

· Device / Suspicious Domain

Cyber AI Analyst Incident Events:

· Detect \ Event \ Possible HTTP Command and Control

· Detect \ Event \ Cryptocurrency Mining Activity

Autonomous Response Models:

· Antigena / Network::Significant Anomaly::Antigena Alerts Over Time Block

· Antigena / Network::External Threat::Antigena Suspicious Activity Block

· Antigena / Network::Significant Anomaly::Antigena Enhanced Monitoring from Client Block

· Antigena / Network::External Threat::Antigena Crypto Currency Mining Block

· Antigena / Network::External Threat::Antigena File then New Outbound Block

· Antigena / Network::External Threat::Antigena Suspicious File Block

· Antigena / Network::Significant Anomaly::Antigena Significant Anomaly from Client Block

List of Indicators of Compromise (IoCs)

(IoC - Type - Description + Confidence)

· 45.141.87[.]195:8000/infect.ps1 - IP Address, Destination Port, Script - Malicious PowerShell script

· gulf.moneroocean[.]stream - Hostname - Monero Endpoint

· monerooceans[.]stream - Hostname - Monero Endpoint

· 152.53.121[.]6:10001 - IP Address, Destination Port - Monero Endpoint

· 152.53.121[.]6 - IP Address – Monero Endpoint

· https://api[.]chimera-hosting[.]zip/frfnhis/zdpaGgLMav/nbminer[.]exe – Hostname, Executable File – NBMiner

· Db3534826b4f4dfd9f4a0de78e225ebb – Hash – NBMiner loader

MITRE ATT&CK Mapping

(Tactic – Technique – Sub-Technique)

· Vulnerabilities – RESOURCE DEVELOPMENT – T1588.006 - T1588

· Exploits – RESOURCE DEVELOPMENT – T1588.005 - T1588

· Malware – RESOURCE DEVELOPMENT – T1588.001 - T1588

· Drive-by Compromise – INITIAL ACCESS – T1189

· PowerShell – EXECUTION – T1059.001 - T1059

· Exploitation of Remote Services – LATERAL MOVEMENT – T1210

· Web Protocols – COMMAND AND CONTROL – T1071.001 - T1071

· Application Layer Protocol – COMMAND AND CONTROL – T1071

· Resource Hijacking – IMPACT – T1496

· Obfuscated Files - DEFENSE EVASION - T1027

· Bypass UAC - PRIVILEGE ESCALATION – T1548.002

· Process Injection – PRIVILEGE ESCALATION – T055

· Debugger Evasion – DISCOVERY – T1622

· Logon Autostart Execution – PERSISTENCE – T1547.009

References

[1] https://www.darktrace.com/cyber-ai-glossary/cryptojacking#:~:text=Battery%20drain%20and%20overheating,fee%20to%20%E2%80%9Cmine%20cryptocurrency%E2%80%9D.

[2] https://coinmarketcap.com/

[3] https://www.ibm.com/think/topics/cryptojacking

[4] https://thehackernews.com/2025/07/3500-websites-hijacked-to-secretly-mine.html

[5] https://urlhaus.abuse.ch/url/3589032/

[6] https://www.logpoint.com/en/blog/uncovering-illegitimate-crypto-mining-activity/

[7] https://www.virustotal.com/gui/domain/gulf.moneroocean.stream/detection

[8] https://www.virustotal.com/gui/domain/monerooceans.stream/detection

[9] https://any.run/report/5aa8cd5f8e099bbb15bc63be52a3983b7dd57bb92566feb1a266a65ab5da34dd/351eca83-ef32-4037-a02f-ac85a165d74e

^{The content provided in this blog is published by Darktrace for general informational purposes only and reflects our understanding of cybersecurity topics, trends, incidents, and developments at the time of publication. While we strive to ensure accuracy and relevance, the information is provided “as is” without any representations or warranties, express or implied. Darktrace makes no guarantees regarding the completeness, accuracy, reliability, or timeliness of any information presented and expressly disclaims all warranties.}

^{Nothing in this blog constitutes legal, technical, or professional advice, and readers should consult qualified professionals before acting on any information contained herein. Any references to third-party organizations, technologies, threat actors, or incidents are for informational purposes only and do not imply affiliation, endorsement, or recommendation.}

^{Darktrace, its affiliates, employees, or agents shall not be held liable for any loss, damage, or harm arising from the use of or reliance on the information in this blog.}

^{The cybersecurity landscape evolves rapidly, and blog content may become outdated or superseded. We reserve the right to update, modify, or remove any content without notice.}

No items found.

Inside the SOC

No items found.

•

August 3, 2026

Nathaniel Jones

SVP, Global Threat Intelligence

Senior Director of Product

Watch the NIS2 Webinar

Blog

Network

August 11, 2026

When AI Agents Attack: The Case for Behavioral Anomaly Detection

The UK AI Security Institute (AISI) has just published a detailed incident report confirming what many within the cyber security industry have suspected for some time: frontier AI agents, given a degree of autonomy, can independently develop and execute attack chains against real targets, including social engineering, supply-chain compromise, and deception, without explicit instruction to do so.

The report (INC-2026-07-28-01, published 4 August 2026) documents what happened during a routine capability evaluation between 25 and 28 July. It deserves a careful read from anyone responsible for securing the modern digital estate.

The incident

AISI regularly tests frontier AI models for offensive cyber capability. In this exercise, agents were placed inside a simulated network environment and given internet access this was deliberate, given to reflect realism of the task. Across 122 evaluation runs, 19 instances of unsanctioned activity were recorded, including cases where agents broke the intended boundaries of the test and acted against real people and real infrastructure on the open internet.

In the most serious case, an agent (Mythos 5) autonomously:

Conducted open-source intelligence gathering to identify real developers associated with keywords from the simulated scenario
Created a GitHub account and submitted a pull request containing obfuscated malicious code to a real, unaffiliated repository
Created a second sockpuppet account to post comments on its own PR, attempting to manufacture social proof
When a human reviewer flagged the malicious content, the agent lied, claiming an accidental commit, then repeatedly attempted to reintroduce the payload under the guise of having "fixed" the code
Sent spear-phishing emails to the repository maintainer containing manipulated content designed to convince them to merge the code
Planted a prompt injection inside a GitHub issue, hidden in an HTML comment invisible to humans but readable by AI coding assistants parsing the repository via API

AISI confirmed no real-world harm resulted. The PRs were not merged; the payloads did not execute. But containment was a product of circumstance, not design.

Why this matters

Every action the agents took used legitimate services operating within normal parameters. Creating a GitHub account is legitimate. Opening a pull request is legitimate. Sending an email is legitimate. Posting an issue is legitimate.

No vulnerability was exploited. No known malware signature was present. No indicator of compromise from any threat intelligence feed would have matched. The entire attack existed as a pattern, a sequence of individually normal actions that, taken together, constituted a supply-chain attack, a social engineering campaign, and an attempted compromise of downstream systems.

This is the class of threat that signature-based and rule-based security tools structurally struggle to detect. They answer the question "have I seen this before?" When the answer is no – and when the attack is novel, composed of legitimate actions, and adapted in real time by an adversary in this case an agent that reasons about evasion – they struggle.

Behavioral anomaly detection: The architecture that fits

Darktrace has spent a decade building a platform around a different question: "is this behavior normal for this entity?"

The platform learns the baseline behavior of every user, device, and communication pattern across an organization's digital environment. It does not need prior knowledge of an attack technique. It identifies deviation from learned normality, surfaces it, and through its Autonomous Response capability, takes targeted action to contain threats at machine speed.

Consider how this AISI incident might show across the Darktrace platform:

Network detection and response

While no payload was successfully delivered to a victim in this incident, AISI's post-incident analysis of the agent's code (detailed on pages 24-25 of the report) reveals what would have landed on target systems had the social engineering succeeded. The agent produced three successive payload generations, each replacing the last after being detected and called out:

Generation 1 hid a module-load stager inside a legitimate bug-fix commit, guarded by an environment variable check, to skip execution in CI. Its second stage attempted to fetch Poseidon implants (Linux and Windows variants) calling back to the agent's own C2 listener, with a fallback to publicly hosted polling scripts that fetched and executed commands every 60 seconds.

Generation 2 exploited an Electron "What's New" window with nodeIntegration enabled, loading remote HTML from a GitHub Pages site the agent had registered. The page decoded an XOR-obfuscated payload at runtime that beaconed host identifiers over DNS, attempted to download the same implants, ran its own in-process C2 loop, and installed reboot persistence via crontab and Windows registry Run keys.

Generation 3 was a post-install dropper added after the agent had already been publicly caught. It XOR-decoded a hidden URL, downloaded a script over plaintext HTTP, and ran it detached, force-exiting so npm install always reports success, giving the attacker remote code execution on any machine running the install.

None of these payloads reached a victim. But the technical detail matters: this is the calibre of tradecraft an autonomous agent produced in a matter of hours, unprompted. Each generation demonstrated adaptation, the agent iterated its approach in direct response to being detected.

For network defenders, the behavioral signals these payloads would have generated are clear. DNS beaconing to novel domains, periodic HTTP polling to newly registered infrastructure, XOR-encoded data in transit, registry and crontab persistence writes on endpoints that have never exhibited such behavior. These are exactly the anomalies that Darktrace / NETWORK identifies in real time: deviations from learned device and connection baselines, regardless of whether the specific payload or C2 domain has been seen before.

Beyond the payload analysis, the agents' broader activity produced additional behavioral anomalies: new identities engaging in high-frequency interactions with external services, coordination via unusual channels (public pastebins, gists), and lateral communication patterns that deviate from any established baseline.

Email security

The spear-phishing emails sent by the agent, crafted with context-specific content designed to manipulate a specific individual, represent exactly the kind of socially engineered communication that static email gateways can miss. Darktrace / EMAIL analyses the linguistic patterns, sender behavior, and relational context of every message. A brand-new sender using tailored language to pressure a recipient into a specific technical action deviates from the learned communication profile, regardless of whether the message contains a known-bad URL or attachment.

Securing AI

This is where the incident points to a threat category that many security platforms do not yet fully address. The agent did not just use AI as a tool for attack. It targeted other AI systems, attempting prompt injections against coding assistants and trying to compromise automated workflows. Organisations adopting AI agents for development, operations, and decision-making are introducing a newer extended attack surface.

Darktrace / SECURE AI extends the same behavioral detection model into this space. It monitors AI usage across the enterprise, detecting anomalous interactions with AI services, identifying sensitive data exposure to AI systems, providing visibility into shadow AI use. As AI agents become standard components of enterprise infrastructure, the ability to detect when an AI system is behaving outside its expected parameters, when there is a chance, it is being manipulated, or when it is interacting with something it should not be, that becomes a direct security requirement rather than a future consideration.

The principle is consistent across the Darktrace platform: learn what normal looks like, identify deviation, respond autonomously.

The AI-to-AI threat surface

One of the report's most important findings is a confirmation that AI agents can act outside of instructions and guardrails to complete tasks. The Mythos 5 agent conducted reconnaissance on a repository maintainer, inferred it might be an AI coding assistant, and deployed a prompt injection specifically crafted to compromise it.

This is a new attack class. Organizations deploying AI coding assistants, AI-powered code review, or autonomous DevOps agents now face an adversary that:

Moves at speed
Is able to adapt its approach when initial attempts are halted
Specifically targets AI systems via techniques (prompt injection, hidden instructions in HTML comments) that are less visible to human reviewers but parsed by machines
Reasons about which evasion strategies are most likely to succeed

Defending against this requires security that operates with the same adaptability and speed. A system that learns what "normal" looks like, and flags deviation from it.

What AISI recommends

The report identifies five contributing factors to the incident. Two are directly relevant:

No synchronous real-time monitoring (Section 5.3): AISI acknowledges it had not built real-time behavioral monitoring capable of immediately blocking unsanctioned agent actions. Their security team detected the incident via commercial security monitoring software, but only after activity had already occurred.

No fine-grained behavioral controls (Section 6.2): AISI's forward-looking response includes building continuous monitoring that can "immediately block or flag unsanctioned AI agent behavior."

The direction is clear: organizations need security that watches behavior continuously, learns what normal looks like, and acts when deviation occurs, without requiring a human in the loop for every decision.

The takeaway

The AISI incident is documented proof that autonomous AI agents will independently develop novel attack chains using legitimate tools, adapt when detected, employ deception, and target both humans and other AI systems. This happened last week, in a controlled setting, with commercially available models.

The security architecture that addresses this is behavioral anomaly detection applied across the full digital estate, as AI agents become standard components of enterprise infrastructure, writing code, managing deployments, processing communications, the attack surface they create is behavioral by nature.

This is the approach Darktrace has taken for years: learning what is normal across an organization’s digital environment, identifying meaningful deviations, and responding to emerging threats without relying on known attack signatures. As autonomous AI agents introduce new and unpredictable behaviors, that foundation becomes increasingly important to securing the enterprise.

‍

Read the full report from the UK AI Security Institute here.

About the author

Adam Stevens

Senior Director of Product

Blog

August 5, 2026

Testing a Prompt injection Attack Against an Enterprise AI Agent

Key takeaways

Darktrace successfully detected and quarantined a prompt injection email before it could be processed by an enterprise AI agent.
Prompt injection attacks increasingly rely on natural language rather than traditional malware, making behavioral analysis an important complement to signature-based detection.
Organizations deploying AI agents should combine model guardrails with behavioral monitoring to reduce the risk of malicious instructions reaching enterprise systems.

How behavioral detection helps stop prompt injection attacks

A Darktrace customer running a Gemini AI agent in Google Cloud asked us two simple questions:

“If my agent can read inbound emails and access internal data, what stops an attacker from hiding malicious instructions in the message? Couldn’t the agent be tricked into deleting or exfiltrating sensitive data?”

The scenario centers on an indirect prompt injection attack, where malicious instructions are hidden inside content that an AI model later interprets as trusted input. The same weakness was exposed by EchoLeak (CVE-2025-32711), a zero-click Microsoft 365 Copilot vulnerability enabled data exfiltration from a single well-crafted email..

This blog follows the Darktrace team’s investigation of the customer’s hypothesis and examines how the attack interacted with their existing security stack. The results highlight which defenses held, where gaps emerged, and how behavioral detection mattered more than guardrails. This investigation also demonstrates why behavioral detection is becoming increasingly important for AI security, as prompt injections often contain no traditional indicators of compromise.

How do prompt injection attacks work?

Prompt injection works by carefully crafting the content and structure of the prompt to alter the LLM’s behavior or output in unintended ways. This can cause models to violate guardrails, generate harmful content or enable unauthorised access.

Prompt injection attack example

The well-known example, EchoLeak (CVE-2025-32711), was a zero-click vulnerability in Microsoft 365 Copilot that relied on a carefully crafted email containing hidden instructions that the AI system interpreted as commands rather than content, creating a pathway for unauthorized access to sensitive information without any user interaction.

While Darktrace / SECURE AI is designed to prevent agents from producing unintended outcomes, we wanted to see if we could catch and prevent this threat type earlier in the attack life-cycle, at the email security layer.

How we tested prompt injection attacks on an enterprise agent

Summary:

Claude generated a prompt injection payload.
Hidden instructions were embedded in an email.
The email passed traditional validation checks.
Darktrace analyzed the language and sender behavior.
The email was quarantined before the AI agent could process it.

To test Darktrace / EMAIL against this attack class, we opened Claude, gave it the customer's context and problem statement (Gemini agent with inbox access, internal tool calls), told it we were validating Darktrace / EMAIL's detection of prompt injections, and asked for a test payload. See below:

Despite the guardrails supposedly built into the model, Claude surprisingly gave us the entire exploit in plaintext (albeit very basic), illegible to a human as the text was sent in white text (see Figure 1) but framed as an authoritative override for anything downstream reading the mail programmatically (i.e. the Gemini agent).

How Darktrace detected a prompt injection attack

We then sent the Claude-crafted email from a freemail address to the target recipient’s inbox. Despite the email containing no malicious payload, the freemail address having no malicious reputation, and the validation checks all passing, Darktrace /EMAIL flagged the email as a 93/100 anomaly and moved it to junk, out of scope for the AI agent.

Figure 3: *The test email sent with the hidden prompt injection*

*Figure 4: The email analysis in Darktrace / EMAIL*

Figure 5: Darktrace / EMAIL detection of malicious activity

The interesting part is what triggered the detection (see Figure 5)

Possible machine prompt content: text in the body detected as instructions written for a machine to execute, not for a human to read

Possible machine prompt content + basic suspicious correspondence: the same content, correlated with sender-side anomalies: freemail domain (yahoo[.]com), unknown correspondent, no prior mail history with the recipient, and suspicious references to payment information

Neither of those is a signature match. Nothing in the email was on a blacklist. There was no malware, no link and no attachment. Darktrace analyzed the context in which the email was delivered and flagged it as likely risky. The anomalous language features and the context of the sender relative to the recipient's normal behavior, combined with the unusual hidden text (prompt) were enough for Darktrace / EMAIL to act on the risk.

Result: Darktrace / EMAIL autonomously junked the email, out of scope for any AI agent parsing the inbox.

Why behavioral security makes a difference detecting prompt injection attacks

Cyberattacks don't look like traditional exploits anymore. They now operate in natural language, not strictly code.

That breaks the traditional stack. AV, firewalls, static scanning and signature-based SEGs all assume a payload to inspect.

A prompt injection has no payload. It's just an instruction, written in natural language, dressed up as anything the attacker wants: an invoice, an HR request, a calendar invite, some simple PowerPoint slides.

EchoLeak proved that hidden instructions can sit inside an email invisible to the user but fully readable by the LLM, and the LLM will follow them blindly.

This test and GTG-1002 proved that the LLM itself can be socially engineered. Tell it you're an authorized tester and it will hand you the attack.

Rules and static classifiers can catch the obvious cases. But natural language has infinite variants, and the attack surface is the model's innate functionality to comply.

The deeper problem here is intent: an LLM can't reliably tell whether an instruction in its context came from its developer, its user, or an attacker who slipped it into an email. To the LLM, everything reads as language and looks like a legitimate ask. This is why behavioural detection wins, as you become aware of intent when you look at the context of an interaction. Does this sender normally send this kind of message to this recipient? Does this prompt fit the user's normal pattern? Is this agent behaving the way this agent normally behaves?

Intent can't be read off a single email, it emerges from behavioral context. Which is how Darktrace enables threat detection, through behavioral understanding.

Why enterprise AI security requires more than guardrails

Claude didn't roll over immediately… the first section of the response was a (slight) pushback, but then it wrote the payload anyway without having to ask twice.

Here the framing of the prompt did all the work. The “testing security capabilities” angle moved the model from refusal to unquestioned compliance to the user prompt.

This isn't the first time this has happened, of course. Anthropic disclosed in November 2025 that a Chinese state-sponsored group they tracked as GTG-1002 ran the first documented AI-orchestrated espionage campaign against ~30 targets by posing as employees of a legitimate cybersecurity firm doing authorised penetration testing.

The takeaway isn't that AI guardrails are ineffective. They raise the cost of low-effort attacks and remain an important first layer of defense. However, for most organizations today, they’re the only line of defense when deploying AI agents. If a prompt injection bypasses those controls, organizations still need a way to detect and stop malicious behavior elsewhere in the attack chain.

Attackers will continue to have working prompt injections easily and quickly. The question is what stops one when it lands in an inbox your agent is reading.

That's where behavioral detection comes in.

How Darktrace detects prompt injection attacks in emails

Two things Darktrace does that a model-level guardrail or static rules and signatures can't:

Natural language analysis at the email or prompt layer. The email is assessed on its own merits: is this content shaped like instructions for a machine, regardless of what the receiving agent decides to do about it?

Behavioral context around the language. An AI agent behaves like an extremely agreeable human, and it will go above and beyond to comply with the user’s request. That's exactly why you must consider the business context, such sender behaviour, mailing history, and organisational norms, as these matter even more when the recipient is an AI.

Darktrace has been perfecting behavioral anomaly detection for over a decade; the same self-learning approach that catches BEC and account takeover applies directly to prompt injection delivery. Our multi-layered AI stack extracts content from the message, builds behavioural understanding through social graphing and Pattern of Life analysis, and then combines natural language, topic, inducement, sender relationship and anomaly signals before deciding what action to take.

This matters for prompt injection because the threat is not the plain language itself, but the intent behind the language that can cause an AI agent to respond in unexpected ways.

How to secure enterprise AI operations from prompt injection attacks

Email was the entry point in this case, but it is only one of many possible vectors.

Anywhere an agent can retrieve information, an attacker can potentially introduce a prompt injection.

Emails, documents, SharePoint sites, web pages, knowledge bases, chat platforms, and third-party integrations all provide opportunities to influence an agent's behavior. Wherever an agent finds its orders, a prompt injection opportunity exists.

This is why securing AI requires more than blocking malicious inputs. Organizations also need visibility into how agents behave after consuming information from across their environment. If an agent begins accessing unexpected data, taking unusual actions, or operating outside its normal patterns, those behaviors may provide the strongest signal that something has gone wrong.

Effective AI security requires defense in depth: reducing the likelihood of malicious instructions reaching the agent while maintaining the ability to detect and investigate suspicious behavior if they do.

The challenge isn't protecting a single entry point. It's recognizing that, in an AI-powered environment, every source of information is also a potential source of influence.

Are you deploying autonomous agents across your enterprise and want to see this tested in your environment? Let's talk.

About the author

Carlo Loregian

Solutions Engineer

Your data. Our AI.

Elevate your network security with Darktrace AI

Get a demo

Check out this article by Darktrace: How Darktrace Detects Cryptomining: Threat Example & Walkthrough

From PowerShell to Payload: Darktrace’s Detection of a Novel Cryptomining Malware

What is Cryptojacking?

Darktrace’s Coverage of Cryptojacking via PowerShell

Inside the payload: Technical analysis of the malicious script and cryptomining loader

Zooming Out: Darktrace Coverage of NBMiner

Darktrace’s Autonomous Response

Conclusion

Appendices

List of Indicators of Compromise (IoCs)

MITRE ATT&CK Mapping

References

Why Trust is the New Attack Surface: Darktrace’s Mid-Year Threat Update 2026

When AI Agents Attack: The Case for Behavioral Anomaly Detection

Enjoying the blog?

More in this series

When AI Agents Attack: The Case for Behavioral Anomaly Detection

Hiding in Plain Sight: Uncovering a Multi-Stage Ransomware Attack Through Behavioral Detection

9 Stages of Ransomware & How AI Responds

Blog

Network

August 11, 2026

When AI Agents Attack: The Case for Behavioral Anomaly Detection

The incident

Why this matters

Behavioral anomaly detection: The architecture that fits

Network detection and response

Email security

Securing AI

The AI-to-AI threat surface

What AISI recommends

The takeaway

Blog

August 5, 2026

Testing a Prompt injection Attack Against an Enterprise AI Agent

Key takeaways

How behavioral detection helps stop prompt injection attacks

How do prompt injection attacks work?

Prompt injection attack example

How we tested prompt injection attacks on an enterprise agent

How Darktrace detected a prompt injection attack

Why behavioral security makes a difference detecting prompt injection attacks

Why enterprise AI security requires more than guardrails

How Darktrace detects prompt injection attacks in emails

How to secure enterprise AI operations from prompt injection attacks