Blog
/
Network
/
May 2, 2025

SocGholish: From loader and C2 activity to RansomHub deployment

In early 2025, Darktrace uncovered SocGholish-to-RansomHub intrusion chains, including loader and C2 activity, alongside credential harvesting via WebDAV and SCF abuse. Learn more about SocGholish and its kill chain here!
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
Christina Kreza
Cyber Analyst
laptop and hand typingDefault blog image
02
May 2025

Over the past year, a clear pattern has emerged across the threat landscape: ransomware operations are increasingly relying on compartmentalized affiliate models. In these models, initial access brokers (IABs) [6], malware loaders, and post-exploitation operators work together.

Due to those specialization roles, a new generation of loader campaigns has risen. Threat actors increasingly employ loader operators to quietly establish footholds on the target network. These entities then hand off access to ransomware affiliates. One loader that continues to feature prominently in such campaigns is SocGholish.

What is SocGholish?

SocGholish is a loader malware that has been utilized since at least 2017 [7].  It has long been associated with fake browser updates and JavaScript-based delivery methods on infected websites.

Threat actors often target outdated or poorly secured CMS-based websites like WordPress. Through unpatched plugins, or even remote code execution flaws, they inject malicious JavaScript into the site’s HTML, templates or external JS resources [8].  Historically, SocGholish has functioned as a first-stage malware loader, ultimately leading to deployment of Cobalt Strike beacons [9], and further facilitating access persistence to corporate environments. More recently, multiple security vendors have reported that infections involving SocGholish frequently lead to the deployment of RansomHub ransomware [3] [5].

This blog explores multiple instances within Darktrace's customer base where SocGholish deployment led to subsequent network compromises. Investigations revealed indicators of compromise (IoCs) similar to those identified by external security researchers, along with variations in attacker behavior post-deployment. Key innovations in post-compromise activities include credential access tactics targeting authentication mechanisms, particularly through the abuse of legacy protocols like WebDAV and SCF file interactions over SMB.

Initial access and execution

Since January 2025, Darktrace’s Threat Research team observed multiple cases in which threat actors leveraged the SocGholish loader for initial access. Malicious actors commonly deliver SocGholish by compromising legitimate websites by injecting malicious scripts into the HTML of the affected site. When the visitor lands on an infected site, they are typically redirected to a fake browser update page, tricking them into downloading a ZIP file containing a JavaScript-based loader [1] [2]. In one case, a targeted user appears to have visited the compromised website garagebevents[.]com (IP: 35.203.175[.]30), from which around 10 MB of data was downloaded.

Device Event Log showing connections to the compromised website, following by connections to the identified Keitaro TDS instances.
Figure 1: Device Event Log showing connections to the compromised website, following by connections to the identified Keitaro TDS instances.

Within milliseconds of the connection establishment, the user’s device initiated several HTTPS sessions over the destination port 443 to the external endpoint 176.53.147[.]97, linked to the following Keitaro TDS domains:

  • packedbrick[.]com
  • rednosehorse[.]com
  • blackshelter[.]org
  • blacksaltys[.]com

To evade detection, SocGholish uses highly obfuscated code and relies on traffic distribution systems (TDS) [3].  TDS is a tool used in digital and affiliate marketing to manage and distribute incoming web traffic based on predefined rules. More specifically, Keitaro is a premium self-hosted TDS frequently utilized by attackers as a payload repository for malicious scripts following redirects from compromised sites. In the previously noted example, it appears that the device connected to the compromised website, which then retrieved JavaScript code from the aforementioned Keitaro TDS domains. The script served by those instances led to connections to the endpoint virtual.urban-orthodontics[.]com (IP: 185.76.79[.]50), successfully completing SocGholish’s distribution.

Advanced Search showing connections to the compromised website, following by those to the identified Keitaro TDS instances.
Figure 2: Advanced Search showing connections to the compromised website, following by those to the identified Keitaro TDS instances.

Persistence

During some investigations, Darktrace researchers observed compromised devices initiating HTTPS connections to the endpoint files.pythonhosted[.]org (IP: 151.101.1[.]223), suggesting Python package downloads. External researchers have previously noted how attackers use Python-based backdoors to maintain access on compromised endpoints following initial access via SocGholish [5].

Credential access and lateral movement

Credential access – external

Darktrace researchers identified observed some variation in kill chain activities following initial access and foothold establishment. For example, Darktrace detected interesting variations in credential access techniques. In one such case, an affected device attempted to contact the rare external endpoint 161.35.56[.]33 using the Web Distributed Authoring and Versioning (WebDAV) protocol. WebDAV is an extension of the HTTP protocol that allows users to collaboratively edit and manage files on remote web servers. WebDAV enables remote shares to be mounted over HTTP or HTTPS, similar to how SMB operates, but using web-based protocols. Windows supports WebDAV natively, which means a UNC path pointing to an HTTP or HTTPS resource can trigger system-level behavior such as authentication.

In this specific case, the system initiated outbound connections using the ‘Microsoft-WebDAV-MiniRedir/10.0.19045’ user-agent, targeting the URI path of /s on the external endpoint 161.35.56[.]33. During these requests, the host attempted to initiate NTML authentication and even SMB sessions over the web, both of which failed. Despite the session failures, these attempts also indicate a form of forced authentication. Forced authentication exploits a default behavior in Windows where, upon encountering a UNC path, the system will automatically try to authenticate to the resource using NTML – often without any user interaction. Although no files were directly retrieved, the WebDAV server was still likely able to retrieve the user’s NTLM hash during the session establishment requests, which can later be used by the adversary to crack the password offline.

Credential access – internal

In another investigated incident, Darktrace observed a related technique utilized for credential access and lateral movement. This time, the infected host uploaded a file named ‘Thumbs.scf’ to multiple internal SMB network shares. Shell Command File ( SCF) is a legacy Windows file format used primarily for Windows Explorer shortcuts. These files contain instructions for rendering icons or triggering shell commands, and they can be executed implicitly when a user simply opens a folder containing the file – no clicks required.

The ‘Thumbs.scf’ file dropped by the attacker was crafted to exploit this behavior. Its contents included a [Shell] section with the Command=2 directive and an IconFile path pointing to a remote UNC resource on the same external endpoint, 161.35.56[.]33, seen in the previously described case – specifically, ‘\\161.35.56[.]33\share\icon.ico’. When a user on the internal network navigates to the folder containing the SCF file, their system will automatically attempt to load the icon. In doing so, the system issues a request to the specified UNC path, which again prompts Windows to initiate NTML authentication.

This pattern of activity implies that the attacker leveraged passive internal exposure; users who simply browsed a compromised share would unknowingly send their NTML hashes to an external attacker-controlled host. Unlike the WebDAV approach, which required initiating outbound communication from the infected host, this SCF method relies on internal users to interact with poisoned folders.

Figure 3: Contents of the file 'Thumbs.scf' showing the UNC resource hosted on the external endpoint.
Figure 3: Contents of the file 'Thumbs.scf' showing the UNC resource hosted on the external endpoint.

Command-and-control

Following initial compromise, affected devices would then attempt outbound connections using the TLS/SSL protocol over port 443 to different sets of command-and-control (C2) infrastructure associated with SocGholish. The malware frequently uses obfuscated JavaScript loaders to initiate its infection chain, and once dropped, the malware communicates back to its infrastructure over standard web protocols, typically using HTTPS over port 443. However, this set of connections would precede a second set of outbound connections, this time to infrastructure linked to RansomHub affiliates, possibly facilitating the deployed Python-based backdoor.

Connectivity to RansomHub infrastructure relied on defense evasion tactics, such as port-hopping. The idea behind port-hopping is to disguise C2 traffic by avoiding consistent patterns that might be caught by firewalls, and intrusion detection systems. By cycling through ephemeral ports, the malware increases its chances of slipping past basic egress filtering or network monitoring rules that only scrutinize common web traffic ports like 443 or 80. Darktrace analysts identified systems connecting to destination ports such as 2308, 2311, 2313 and more – all on the same destination IP address associated with the RansomHub C2 environment.

Figure 4: Advanced Search connection logs showing connections over destination ports that change rapidly.

Conclusion

Since the beginning of 2025, Darktrace analysts identified a campaign whereby ransomware affiliates leveraged SocGholish to establish network access in victim environments. This activity enabled multiple sets of different post exploitation activity. Credential access played a key role, with affiliates abusing WebDAV and NTML over SMB to trigger authentication attempts. The attackers were also able to plant SCF files internally to expose NTML hashes from users browsing shared folders. These techniques evidently point to deliberate efforts at early lateral movement and foothold expansion before deploying ransomware. As ransomware groups continue to refine their playbooks and work more closely with sophisticated loaders, it becomes critical to track not just who is involved, but how access is being established, expanded, and weaponized.

Credit to Chrisina Kreza (Cyber Analyst) and Adam Potter (Senior Cyber Analyst)

[related-resource]

Appendices

Darktrace / NETWORK model alerts

·       Anomalous Connection / SMB Enumeration

·       Anomalous Connection / Multiple Connections to New External TCP Port

·       Anomalous Connection / Multiple Failed Connections to Rare Endpoint

·       Anomalous Connection / New User Agent to IP Without Hostname

·       Compliance / External Windows Communication

·       Compliance / SMB Drive Write

·       Compromise / Large DNS Volume for Suspicious Domain

·       Compromise / Large Number of Suspicious Failed Connections

·       Device / Anonymous NTML Logins

·       Device / External Network Scan

·       Device / New or Uncommon SMB Named Pipe

·       Device / SMB Lateral Movement

·       Device / Suspicious SMB Activity

·       Unusual Activity / Unusual External Activity

·       User / Kerberos Username Brute Force

MITRE ATT&CK mapping

·       Credential Access – T1187 Forced Authentication

·       Credential Access – T1110 Brute Force

·       Command and Control – T1071.001 Web Protocols

·       Command and Control – T1571 Non-Standard Port

·       Discovery – T1083 File and Directory Discovery

·       Discovery – T1018 Remote System Discovery

·       Discovery – T1046 Network Service Discovery

·       Discovery – T1135 Network Share Discovery

·       Execution – T1059.007 JavaScript

·       Lateral Movement – T1021.002 SMB/Windows Admin Shares

·       Resource Deployment – T1608.004 Drive-By Target

List of indicators of compromise (IoCs)

·       garagebevents[.]com – 35.203.175[.]30 – Possibly compromised website

·       packedbrick[.]com – 176.53.147[.]97 – Keitaro TDS Domains used for SocGholish Delivery

·       rednosehorse[.]com – 176.53.147[.]97 – Keitaro TDS Domains used for SocGholish Delivery

·       blackshelter[.]org – 176.53.147[.]97 – Keitaro TDS Domains used for SocGholish Delivery

·       blacksaltys[.]com – 176.53.147[.]97 – Keitaro TDS Domains used for SocGholish Delivery

·       virtual.urban-orthodontics[.]com – 185.76.79[.]50

·       msbdz.crm.bestintownpro[.]com – 166.88.182[.]126 – SocGholish C2

·       185.174.101[.]240 – RansomHub Python C2

·       185.174.101[.]69 – RansomHub Python C2

·       108.181.182[.]143 – RansomHub Python C2

References

[1] https://www.checkpoint.com/cyber-hub/threat-prevention/what-is-malware/socgholish-malware/

[2] https://intel471.com/blog/threat-hunting-case-study-socgholish

[3] https://www.trendmicro.com/en_us/research/25/c/socgholishs-intrusion-techniques-facilitate-distribution-of-rans.html

[4] https://www.proofpoint.com/us/blog/threat-insight/update-fake-updates-two-new-actors-and-new-mac-malware

[5] https://www.guidepointsecurity.com/blog/ransomhub-affiliate-leverage-python-based-backdoor/

[6] https://www.cybereason.com/blog/how-do-initial-access-brokers-enable-ransomware-attacks

[7] https://attack.mitre.org/software/S1124/

[8] https://expel.com/blog/incident-report-spotting-socgholish-wordpress-injection/

[9] https://www.esentire.com/blog/socgholish-to-cobalt-strike-in-10-minutes

Get the latest insights on emerging cyber threats

This report explores the latest trends shaping the cybersecurity landscape and what defenders need to know in 2026

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
Christina Kreza
Cyber Analyst

More in this series

No items found.

Blog

/

Network

/

August 6, 2026

When AI Agents Attack: The Case for Behavioral Anomaly Detection

Default blog imageDefault blog image

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. The defense must be too.

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

Continue reading
About the author
Adam Stevens
Senior Director of Product | Darktrace

Blog

/

AI

/

August 5, 2026

Testing a Prompt injection Attack Against an Enterprise AI Agent

prompt injectionDefault blog imageDefault blog image

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:

  1. Claude generated a prompt injection payload.  
  2. Hidden instructions were embedded in an email.  
  3. The email passed traditional validation checks.  
  4. Darktrace analyzed the language and sender behavior.  
  5. 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:

Figure 1
Figure 2

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.

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