When AI Becomes the Lure: A Fake Gemini Installer Delivers Vidar

As organizations are increasingly adopting generative AI tools into their daily workflows, attackers are adapting their distribution methods accordingly too. As part of their day-to-day work, users are now searching for AI assistants, programming tools, browser extensions, desktop applications, and productivity integrations.

Recent reports have highlighted campaigns that use fake AI software and AI-related installers to distribute malware and steal credentials [1]. Researchers have documented campaigns that exploit fake AI-themed websites and services to distribute information stealers and backdoors [2]. Security researchers have also observed attackers disguising malware as legitimate installers for AI software to increase the likelihood of victim interaction and execution [3].

In July 2026, Darktrace observed one such case within a customer environment in the Europe, Middle East and Africa (EMEA) region, where attackers used a fake generative AI installer to deliver the prolific information stealer Vidar . This incident highlights how threat actors are exploiting interest in AI services to distribute established malware using increasingly convincing social engineering techniques.

Unlike many malware campaigns that begin with a phishing email, this activity appears to have originated from a user searching for and downloading software.

Darktrace first observed unusual activity on the customer network after a suspicious executable file was launched from a user’s Download folder. Further investigation revealed that the file purported to be a Google Gemini installer and was named “Download_Google_Gemini_For_Windows.exe”.

During the initial analysis, it was noted that the top result for the suspicious filename associated pointed to a file hosted on Google Colab, a cloud-based Jupyter notebook platform, commonly used by developers, researchers, and data scientists to run code and machine learning workloads through a web browser. By leveraging another trusted Google platform, the attacker increased the likelihood that users would perceive the download as legitimate, making the lure more convincing to those searching for Gemini-related software.

Further investigation of the Google Colab page revealed that the download prompt redirected users to a secondary site, hxxps://micronsoftwares[.]com, which posed as a "Windows Software Hub" download page and offered the fake Gemini installer for download.

While the investigation did not uncover any HTTP or file-download telemetry data that conclusively identified the download source, SSL communication sessions with Google Colab were detected immediately before the suspicious file was executed. The timing of these connections suggests that the user interacted with the Colab resource before being redirected to the secondary site from which the executable was downloaded.

The user was not simply tricked into opening an email attachment; instead, the attacker embedded malicious content into a process many users would consider entirely legitimate: searching for and downloading software associated with a trusted platform.

At the time of review (July 15, 2026), Darktrace's Threat Research team confirmed that the Google Colab page was still active and prompting users to download a ZIP archive containing the binary file.

The archive also appeared to contain a README file instructing users to run the binary file with administrator privileges and add it to their antivirus software’s exception lists. These instructions suggest that the campaign relied heavily on social engineering, convincing users to take actions that would facilitate malware execution and potentially bypass security checks.

The use of a legitimate platform also complicates the user’s decision-making. Downloads associated with a trusted service are often perceived as less suspicious than those hosted on unfamiliar domains. When combined with the branding of a widely used AI tool, the lure becomes even more convincing.

Darktrace’s Threat Research team identified the executable file as the information-stealing malware Vidar. Analysis revealed that the binary file was a newer Go-compiled variant that communicated with Telegram-based infrastructure. Darktrace’s researchers also identified dtm[.]kijangturbo88[.]top as a command-and-control (C2) endpoint associated with the activity. While the malware itself was not novel, the lure and delivery mechanism was. For a deeper look at the information stealer, see Darktrace’s 2023 analysis of Vidar .

Shortly after execution, the process established communications with the external IP address 91.98.98[.]86 via port 443, directly linking the executable to suspicious network activity observed on the device. Subsequent open-source intelligence (OSINT) analysis of the revealed multiple malicious associations [5].

Additional Darktrace detections included unusual SSL activity from the affected device. Analysis of related SSL telemetry identified 91.98.111[.]49 as additional infrastructure associated with the activity [6].

Subsequent alerts from the customer's Microsoft Defender for Endpoint integration later confirmed activity consistent with the theft of browser credentials and other sensitive data from the affected endpoint.

Taken together, these detections provided a clear picture of the attack, from the execution of a suspicious file and unusual network connections to indicators of C2 activity and credential theft.

Darktrace's Autonomous Response

Following the detection, Darktrace’s Autonomous Response took immediate containment action, including blocking communication with suspicious external infrastructure, including 91.98.98[.]86, and quarantining the compromised device.

Despite the apparent legitimacy of the activity, with the installer hosted on a trusted platform and resembling a routine software download, Darktrace was able to detect and contain the attack because the device's behavior deviated from its normal pattern.

This investigation highlights how threat actors continue to adapt established malware delivery techniques to emerging technology trends. While the malware itself was not new, the distribution method was. By disguising Vidar as a Google Gemini installer and hosting the malicious content on a trusted platform, the attack aligned its lure with a growing behavioral trend: users actively searching for AI tools and services as part of their day-to-day work.

Although fake installers are not a new phenomenon, the rapid rise of generative AI has created new opportunities for threat actors. Rather than relying solely on traditional delivery methods, attackers can now target users who are actively searching for AI applications. As AI adoption continues to accelerate across enterprise environments, organizations should remain alert to campaigns that exploit this interest through fake applications, malicious websites, manipulated results, the misuse of trusted platforms, and AI-themed social engineering.

Credit to Rushanth Ramanathan (Cyber Analyst) Joanna Ng (Detection Engineer) Edited by Ryan Traill (Content Manager)

•Antigena / Network / Significant Anomaly / Antigena Significant Anomaly from Client Block

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.

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:

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.

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 evasion – they struggle.

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:

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.

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.

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.

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:

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.

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 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 .

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

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.

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.

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 .

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).

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.

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

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.

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.

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- 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.

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 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.

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.