April 5, 2022

Explore Internet-Facing System Vulnerabilities

Read about 2021's top four incidents and how Darktrace's advanced threat detection technology identified and mitigated vulnerabilities. Learn more.

Written by

Sam Lister

Specialist Security Researcher

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

Sam Lister

Specialist Security Researcher

Apr 2022

By virtue of their exposure, Internet-facing systems (i.e., systems which have ports open/exposed to the wider Internet) are particularly susceptible to compromise. Attackers typically compromise Internet-facing systems by exploiting zero-day vulnerabilities in applications they run. During 2021, critical zero-day vulnerabilities in the following applications were publicly disclosed:

Microsoft Exchange
Confluence
GitLab
Log4j-using applications

Internet-facing systems running these applications were consequently heavily targeted by attackers. In this post, we will provide examples of compromises of these systems observed by Darktrace’s SOC team in 2021. As will become clear, successful exploitation of weaknesses in Internet-facing systems inevitably results in such systems doing things which they do not normally do. Rather than focusing on identifying attempts to exploit these weaknesses, Darktrace focuses on identifying the unusual behaviors which inevitably ensue. The purpose of this post is to highlight the effectiveness of this approach.

Exchange server compromise

In January, researchers from the cyber security company DEVCORE reported a series of critical vulnerabilities in Microsoft Exchange which they dubbed ‘ProxyLogon’.^[1] ProxyLogon consists of a server-side request forgery (SSRF) vulnerability (CVE-2021-26855) and a remote code execution (RCE) vulnerability (CVE-2021-27065). Attackers were observed exploiting these vulnerabilities in the wild from as early as January 6.^[2] In April, DEVCORE researchers reported another series of critical vulnerabilities in Microsoft Exchange which they dubbed ‘ProxyShell’.^[3] ProxyShell consists of a pre-authentication path confusion vulnerability (CVE-2021-34473), a privilege elevation vulnerability (CVE-2021-34523), and a post-authentication RCE vulnerability (CVE-2021-31207). Attackers were first observed exploiting these vulnerabilities in the wild in August.^[4] In many cases, attackers exploited the ProxyShell and ProxyLogon vulnerabilities in order to create web shells on the targeted Exchange servers. The presence of these web shells provided attackers with the means to remotely execute commands on the compromised servers.

In early August 2021, by exploiting the ProxyShell vulnerabilities, an attacker gained the rights to remotely execute PowerShell commands on an Internet-facing Exchange server within the network of a US-based transportation company. The attacker subsequently executed a number of PowerShell commands on the server. One of these commands caused the server to make a 28-minute-long SSL connection to a highly unusual external endpoint. Within a couple of hours, the attacker managed to strengthen their foothold within the network by installing AnyDesk and CobaltStrike on several internal devices. In mid-August, the attacker got the devices on which they had installed Cobalt Strike to conduct network reconnaissance and to transfer terabytes of data to the cloud storage service, MEGA. At the end of August, the attacker got the devices on which they had installed AnyDesk to execute Conti ransomware and to spread executable files and script files to further internal devices.

In this example, the attacker’s exploitation of ProxyShell immediately resulted in the Exchange Server making a long SSL connection to an unusual external endpoint. This connection caused the model Device / Long Agent Connection to New Endpoint to breach. The subsequent reconnaissance, lateral movement, C2, external data transfer, and encryption behavior brought about by the attacker were also picked up by Darktrace’s models.

A non-exhaustive list of the models that breached as a result of the behavior brought about by the attacker:

Device / Long Agent Connection to New Endpoint
Device / ICMP Address Scan
Anomalous Connection / SMB Enumeration
Anomalous Server Activity / Outgoing from Server
Compromise / Beacon to Young Endpoint
Anomalous Server Activity / Rare External from Server
Compromise / Fast Beaconing to DGA
Compromise / SSL or HTTP Beacon
Compromise / Sustained SSL or HTTP Increase
Compromise / Beacon for 4 Days
Anomalous Connection / Multiple HTTP POSTs to Rare Hostname
Unusual Activity / Enhanced Unusual External Data Transfer
Anomalous Connection / Data Sent to Rare Domain
Anomalous Connection / Uncommon 1 GiB Outbound
Compliance / SMB Drive Write
Anomalous File / Internal / Additional Extension Appended to SMB File
Anomalous Connection / Suspicious Read Write Ratio
Anomalous Connection / Suspicious Read Write Ratio and Unusual SMB
Anomalous Connection / Sustained MIME Type Conversion
Unusual Activity / Anomalous SMB Move & Write
Unusual Activity / Unusual Internal Data Volume as Client or Server
Device / Suspicious File Writes to Multiple Hidden SMB Shares
Compromise / Ransomware / Suspicious SMB Activity
Anomalous File / Internal / Unusual SMB Script Write
Anomalous File / Internal / Masqueraded Executable SMB Write
Device / SMB Lateral Movement
Device / Multiple Lateral Movement Model Breaches

Confluence server compromise

Atlassian’s Confluence is an application which provides the means for building collaborative, virtual workspaces. In the era of remote working, the value of such an application is undeniable. The public disclosure of a critical remote code execution (RCE) vulnerability (CVE-2021-26084) in Confluence in August 2021 thus provided a prime opportunity for attackers to cause havoc. The vulnerability, which arises from the use of Object-Graph Navigation Language (OGNL) in Confluence’s tag system, provides attackers with the means to remotely execute code on vulnerable Confluence server by sending a crafted HTTP request containing a malicious parameter.^[5] Attackers were first observed exploiting this vulnerability towards the end of August, and in the majority of cases, attackers exploited the vulnerability in order to install crypto-mining tools onto vulnerable servers.^[6]

At the beginning of September 2021, an attacker was observed exploiting CVE-2021-26084 in order to install the crypto-mining tool, XMRig, as well as a shell script, onto an Internet-facing Confluence server within the network of an EMEA-based television and broadcasting company. Within a couple of hours, the attacker installed files associated with the crypto-mining malware, Kinsing, onto the server. The Kinsing-infected server then immediately began to communicate over HTTP with the attacker’s C2 infrastructure. Around the time of this activity, the server was observed using the MinerGate crypto-mining protocol, indicating that the server had begun to mine cryptocurrency.

In this example, the attacker’s exploitation of CVE-2021-26084 immediately resulted in the Confluence server making an HTTP GET request with an unusual user-agent string (one associated with curl in this case) to a rare external IP. This behavior caused the models Device / New User Agent, Anomalous Connection / New User Agent to IP Without Hostname, and Anomalous File / Script from Rare Location to breach. The subsequent file downloads, C2 traffic and crypto-mining activity also resulted in several models breaching.

A non-exhaustive list of the models which breached as a result of the unusual behavior brought about by the attacker:

Device / New User Agent
Anomalous Connection / New User Agent to IP Without Hostname
Anomalous File / Script from Rare Location
Anomalous File / EXE from Rare External Location
Anomalous File / Internet Facing System File Download
Device / Initial Breach Chain Compromise
Anomalous Connection / Posting HTTP to IP Without Hostname
Compliance / Crypto Currency Mining Activity
Compromise / High Priority Crypto Currency Mining
Device / Internet Facing Device with High Priority Alert

GitLab server compromise

GitLab is an application providing services ranging from project planning to source code management. Back in April 2021, a critical RCE vulnerability (CVE-2021-22205) in GitLab was publicly reported by a cyber security researcher via the bug bounty platform, HackerOne.^[7] The vulnerability, which arises from GitLab’s use of ExifTool for removing metadata from image files, ^[8] enables attackers to remotely execute code on vulnerable GitLab servers by uploading specially crafted image files.^[9] Attackers were first observed exploiting CVE-2021-22205 in the wild in June/July.^[10] A surge in exploitations of the vulnerability was observed at the end of October, with attackers exploiting the flaw in order to assemble botnets.^[11] Darktrace observed a significant number of cases in which attackers exploited the vulnerability in order to install crypto-mining tools onto vulnerable GitLab servers.

On October 29, an attacker successfully exploited CVE-2021-22205 on an Internet-facing GitLab server within the network of a UK-based education provider. The organization was trialing Darktrace when this incident occurred. The attacker installed several executable files and shell scripts onto the server by exploiting the vulnerability. The attacker communicated with the compromised server (using unusual ports) for several days, before making the server transfer large volumes of data externally and download the crypto-mining tool, XMRig, as well as the botnet malware, Mirai. The server was consequently observed making connections to the crypto-mining pool, C3Pool.

In this example, the attacker’s exploitation of the vulnerability in GitLab immediately resulted in the server making an HTTP GET request with an unusual user-agent string (one associated with Wget in this case) to a rare external IP. The models Anomalous Connection / New User Agent to IP Without Hostname and Anomalous File / EXE from Rare External Location breached as a result of this behavior. The attacker’s subsequent activity on the server over the next few days resulted in frequent model breaches.

A non-exhaustive list of the models which breached as a result of the attacker’s activity on the server:

Anomalous Connection / New User Agent to IP Without Hostname
Anomalous File / EXE from Rare External Location
Anomalous File / Multiple EXE from Rare External Locations
Anomalous File / Internet Facing Device with High Priority Alert
Anomalous File / Script from Rare Location
Anomalous Connection / Application Protocol on Uncommon Port
Anomalous Connection / Anomalous SSL without SNI to New External
Device / Initial Breach Chain Compromise
Unusual Activity / Unusual External Data to New IPs
Anomalous Server Activity / Outgoing from Server
Device / Large Number of Model Breaches from Critical Network Device
Anomalous Connection / Data Sent to Rare Domain
Compromise / Suspicious File and C2
Unusual Activity / Enhanced Unusual External Data Transfer
Compliance / Crypto Currency Mining Activity
Compliance / High Priority Crypto Currency Mining
Anomalous File / Zip or Gzip from Rare External Location
Compromise / Monero Mining
Device / Internet Facing Device with High Priority Alert
Anomalous Server Activity / Rare External from Server
Compromise / Slow Beaconing Activity To External Rare
Compromise / Beaconing Activity To External Rare
Compromise / HTTP Beaconing to Rare Destination
Compromise / High Volume of Connections with Beacon Score
Anomalous File / Numeric Exe Download

Log4j server compromise

On December 9 2021, a critical RCE vulnerability (dubbed ‘Log4Shell’) in version 2 of Apache’s Log4j was publicly disclosed by researchers at LunaSec.^[12] As a logging library present in potentially millions of Java applications,^[13] Log4j constitutes an obscured, yet ubiquitous feature of the digital world. The vulnerability (CVE-2021-44228), which arises from Log4j’s Java Naming and Directory Interface (JNDI) Lookup feature, enables an attacker to make a vulnerable server download and execute a malicious Java class file. To exploit the vulnerability, all the attacker must do is submit a specially crafted JNDI lookup request to the server. The fact that Log4j is present in so many applications and that the exploitation of this vulnerability is so simple, Log4Shell has been dubbed the ‘most critical vulnerability of the last decade’.^[14] Attackers have been exploiting Log4Shell in the wild since at least December 1.^[15] Since then, attackers have been observed exploiting the vulnerability to install crypto-mining tools, Cobalt Strike, and RATs onto vulnerable servers.^[16]

On December 10, one day after the public disclosure of Log4Shell, an attacker successfully exploited the vulnerability on a vulnerable Internet-facing server within the network of a US-based architecture company. By exploiting the vulnerability, the attacker managed to get the server to download and execute a Java class file named ‘Exploit69ogQNSQYz.class’. Executing the code in this file caused the server to download a shell script file and a file related to the Kinsing crypto-mining malware. The Kinsing-infected server then went on to communicate over HTTP with a C2 server. Since the customer was using the Proactive Threat Notification (PTN) service, they were immediately alerted to this activity, and the server was subsequently quarantined, preventing crypto-mining activity from taking place.

In this example, the attacker’s exploitation of the zero-day vulnerability immediately resulted in the vulnerable server making an HTTP GET request with an unusual user-agent string (one associated with Java in this case) to a rare external IP. The models Anomalous Connection / Callback on Web Facing Device and Anomalous Connection / New User Agent to IP Without Hostname breached as a result of this behavior. The device’s subsequent file downloads and C2 activity caused several Darktrace models to breach.

A non-exhaustive list of the models which breached as a result of the unusual behavior brought about by the attacker:

Anomalous Connection / Callback on Web Facing Device
Anomalous Connection / New User Agent to IP Without Hostname
Anomalous File / Internet Facing System File Download
Anomalous File / Script from Rare External Location
Device / Initial Breach Chain Compromise
Anomalous Connection / Posting HTTP to IP Without Hostname

Round-up

It is inevitable that attackers will attempt to exploit zero-day vulnerabilities in applications running on Internet-facing devices. Whilst identifying these attempts is useful, the fact that attackers regularly exploit new zero-days makes the task of identifying attempts to exploit them akin to a game of whack-a-mole. Whilst it is uncertain which zero-day vulnerability attackers will exploit next, what is certain is that their exploitation of it will bring about unusual behavior. No matter the vulnerability, whether it be a vulnerability in Microsoft Exchange, Confluence, GitLab, or Log4j, Darktrace will identify the unusual behaviors which inevitably result from its exploitation. By identifying unusual behaviors displayed by Internet-facing devices, Darktrace thus makes it almost impossible for attackers to successfully exploit zero-day vulnerabilities without being detected.

For Darktrace customers who want to find out more about detecting potential compromises of internet-facing devices, refer here for an exclusive supplement to this blog.

Thanks to Andy Lawrence for his contributions.

Footnotes

1. https://devco.re/blog/2021/08/06/a-new-attack-surface-on-MS-exchange-part-1-ProxyLogon/

2. https://www.volexity.com/blog/2021/03/02/active-exploitation-of-microsoft-exchange-zero-day-vulnerabilities/

3. https://www.zerodayinitiative.com/blog/2021/8/17/from-pwn2own-2021-a-new-attack-surface-on-microsoft-exchange-proxyshell

4. https://www.rapid7.com/blog/post/2021/08/12/proxyshell-more-widespread-exploitation-of-microsoft-exchange-servers/

5. https://www.kaspersky.co.uk/blog/confluence-server-cve-2021-26084/23376/

6. https://www.bleepingcomputer.com/news/security/atlassian-confluence-flaw-actively-exploited-to-install-cryptominers/

7. https://hackerone.com/reports/1154542

8. https://security.humanativaspa.it/gitlab-ce-cve-2021-22205-in-the-wild/

9.https://about.gitlab.com/releases/2021/04/14/security-release-gitlab-13-10-3-released/

10. https://www.rapid7.com/blog/post/2021/11/01/gitlab-unauthenticated-remote-code-execution-cve-2021-22205-exploited-in-the-wild/

11. https://www.hackmageddon.com/2021/12/16/1-15-november-2021-cyber-attacks-timeline/

12. https://www.lunasec.io/docs/blog/log4j-zero-day/

13. https://www.csoonline.com/article/3644472/apache-log4j-vulnerability-actively-exploited-impacting-millions-of-java-based-apps.html

14. https://www.theguardian.com/technology/2021/dec/10/software-flaw-most-critical-vulnerability-log-4-shell

15. https://www.rapid7.com/blog/post/2021/12/15/the-everypersons-guide-to-log4shell-cve-2021-44228/

16. https://www.microsoft.com/security/blog/2021/12/11/guidance-for-preventing-detecting-and-hunting-for-cve-2021-44228-log4j-2-exploitation/

執筆者

Sam Lister

Specialist Security Researcher

Inside the SOC

Written by

Sam Lister

Specialist Security Researcher

•

July 10, 2026

Angel Arribas Lopez

Associate Principal Cyber Analyst

•

July 13, 2026

Nathaniel Jones

VP, Security & AI Strategy, Field CISO

Watch the NIS2 Webinar

Blog

AI

July 13, 2026

Security After Signatures: Operating in a World of Pre‑CVE Disclosure Exploitation, Collapsed Trust Boundaries, and Autonomous Systems

Three shifts have reshaped what it means to defend an enterprise securely.

First, exploitation often begins before defenders have a Common Vulnerabilities and Exposures (CVE) identifier, a security advisory, or an entry in the Cybersecurity and Infrastructure Security Agency's (CISA) Known Exploited Vulnerabilities (KEV) catalog.

Secondly, the trust boundary has moved beyond the network edge into identities, tokens, APIs, and Software-as-a-Service (SaaS) workflows.

Third, an increasing share of business activity is executed through automation, integrations, and AI agent-like systems that can act faster than teams can verify intent.

If your security model still relies on detecting known bad artefacts, triaging isolated alerts, and waiting for confirmation before acting, you are already behind the threat.

This is not a failure of security teams; it’s a failure of the operating model to keep pace with how the environment has changed.

A SOC built around alerts and signatures assumes that malicious activity will eventually surface as an event. In real incidents, however, the decisive evidence is rarely a single event. Instead, it is a chain of individually explainable actions that only appears malicious once you connect the dots across identity, non-human identity, cloud, email, SaaS, operational technology (OT), and network telemetry.

The defenders succeeding today observe behaviors, link them into sequences, understand what those sequences mean, and contain impact before the full story unfolds. That is the operating model the current threat environment demands.

Exploitation before disclosure

The first shift is the straightforward: the time to exploit has dropped to nearly zero.

In one example, Darktrace observed a sequence of subtle but strategically significant anomalies within a customer environment that later aligned with exploitation of CVE‑2025‑0994 in Trimble Cityworks by likely Chinese-nexus threat actors. Behavioral indicators were visible at least 18 days before public disclosure, with related anomalies emerging 40 to 50 days earlier during the intrusion window.

This case illustrates a familiar pattern: clusters of weak‑signal anomalies combing to form an actionable picture of intrusion long before a CVE is published. Such activity reflects long‑horizon, option‑preserving operator models often associated with mature state‑linked activity.

Figure 1: Darktrace’s detection of malicious exploitation of CVE 2025-0994, later tied to Chinese-nexus threat actors targeting critical national infrastructure (CNI) in the US, weeks before public disclosure.

Throughout 2025 and 2026, Darktrace has continued to observe the value of anomaly-based detections across a range of incidents.

CVE	CVE Public Disclosure Date	Darktrace Detection Date	Days Between Detection of Exploitation and CVE Public Disclosure
CVE 2025 0994 (Trimble City Works)	2025-02-06	2025-01-19	18 Days
CVE 2025-24183 (Apache)	2025-03-10	2025-02-18	20 days
CVE 2025-10035 (Fortra GoAnywhere)	2025-09-18	2025-09-11	7 days

Identity is the real control plane

The second shift is that identity has replaced perimeter as the primary control plane. As Darktrace’s Annual Threat Report 2026 illustrated, identity remains the main challenge in defending against modern intrusions. A clear example is the Adversary-in-the-Middle (AiTM) case published by Darktrace in December 2025. A phishing email led to the compromise of an Office 365 account. Session hijacking bypassed multi-factor authentication (MFA), and the compromised account was used for follow-on phishing and persistence activities including the creation of malicious email rules.

Every step in that sequence mattered. A successful login alone does not prove legitimacy. An inbox rule, on its own, may not appear catastrophic. Mail activity, viewed in isolation, may seem operationally normal. But the behavioral chain tells a different story: credential theft, token abuse, persistence, and onward compromise through a trusted identity.

This is why the question is no longer “Did the user authenticate successfully”. The more important question is, “Does this identity action make sense right now, in this context, given what came before it?” The AiTM case shows how identity can be compromised. In practice, however, attacks rarely remained confined to identity alone.

In another Darktrace case, a compromised SaaS account triggered activity across the email, SaaS, and network layers, including inbox rule changes, phishing propagation, and connections to suspicious infrastructure. Viewed in isolation, none of these events were decisive. Together, however, they formed a behavioral sequence that revealed the intrusion, with the full attack story automatically correlated and surfaced to defenders by Darktrace’s Cyber AI Analyst.

Figure 2: Cyber AI Analyst correlated and appended additional events to the incident, including other users who connected to the suspicious redirect link after outbound phishing emails were sent.

AI accelerates the threat

The third shift is the one many teams still underestimate: trusted tooling, integrations, and AI agent-like systems can create actions that appear legitimate but are strategically dangerous.

The shift becomes clearer when examining how governments are now framing AI risk. In 2026, guidance published by CISA, UK’s National Cyber Security Centre (NCSC) and Five Eyes partners warned that agentic systems expand attack surfaces, accumulate privilege, and can behave in ways that are difficult to predict or explain [1]. The advice is simple: assume unexpected behavior and design controls around it.

The real risk is not AI usage. It is unknown autonomy: systems with credentials, data access, and action paths that can execute workflow steps without sufficient behavioral validation, traceability, or human oversight. Darktrace’s Model Context Protocol (MCP) risk analysis provides a useful framework for understanding this challenge. Over-privileged agents, content injection, and tool abuse become high-consequence risks when connected systems can dynamically retrieve data, execute actions, and communicate externally.

Whether security teams like it or not, AI is already in the enterprise. It will help drive innovation, but it will also be abused, whether accidentally or maliciously. In each of the cases below, AI either scaled the attacker, built the tooling, or existed within the environment as something to exploit or misuse.

1. AI as an Attack Multiplier

In one campaign targeting Mexican government entities, a single operator used commercial AI platforms to generate exploits, automate reconnaissance, and process large volumes of data, compressing work that would traditionally have required an entire team into a single workflow [2].

Darktrace is also observing this trend further down the stack. In one case, Darktrace identified AI-generated malware exploiting React2Shell, where an attacker used a Large Language Model (LLM) to produce working exploit code and deploy it at scale.

[darktrace.com], [darktrace.com] ‍

2. AI as an Attack Surface

Attempted AI exploitation is now appearing within customer environments. In one case involving an automation technology manufacturer, a compromised LLM proxy was seemingly used as a stepping stone to access additional AI services. When that attempt failed, the attacker pivoted to cryptomining.

What is clear is that the AI layer has already become an asset worth probing, exploiting, and pivoting through. It is also clear that defenders benefit from rapidly understanding how these activities connect. In this case, Cyber AI Analyst automatically pieced together the intrusion, while Darktrace’s Managed Threat Detection service alerted to the customer, enabling the activity to be contained before it could progress further.

Figure 3: Cyber AI Analyst's investigation into a compromised LLM proxy that was abused for cryptomining activity.

AI as a trusted but dangerous actor

This does not require a cinematic vision of “rogue AI.” The Salesloft incident provides a more grounded example, where AI and automation operate with legitimate access but served malicious intent. In that case, attackers abused compromised OAuth tokens associated with the Drift AI chat agent to export significant volumes of data from Salesforce environments.

The activity resembled legitimate API usage and relied on trusted SaaS integrations rather than malware or other obvious signs of intrusion. That is precisely the challenge. Traditional security controls are good at detecting forced entry, but far less effective when a trusted application integration behaves in a way that is technically permitted yet operationally harmful.

In these scenarios, the security challenge shifts from validating access to validating behavior.

This is what that looks like in practice: AI-linked identities executing legitimate actions that require behavioral validation rather than access validation.

Figure 4: Darktrace / SECURE AI highlights anomalous activity across AI identities, surfacing critical behavior that requires validation and containment.

Early observations from Darktrace / SECURE AI deployments reinforce this reality. Across Darktrace's observed fleet, AI service connections per deployment increased 13% during the first half of 2026, reaching over 16 million connections overall. The typical organisation now interacts with seven different AI providers, evidence that AI is no longer operating at the edges of the enterprise. It is increasingly woven into day-to-day business activity.

The most common risks are not compromised models or advanced AI attacks. Instead, they stem from employees and business functions exposing sensitive information through entirely legitimate-looking interactions. Darktrace has observed repeated submission of personally identifiable information (PII), tax information, identification documents, and medical data into LLM prompts, alongside widespread use of unsanctioned (shadow) AI services and growing AI activity from mobile devices.

For defenders, the challenge is increasingly one of context: understanding when legitimate business use crosses into material risk, while preserving privacy and user trust.

Conclusion

Across all three shifts, the pattern is the same: behavior precedes understanding. Security teams are not losing because adversaries have become invisible. An increasingly outdated security model assumes that malicious activity will reveal itself cleanly and early. It no longer does.

In 2026 and beyond, defenders win by understanding behavioral sequences, continuously validating trust, and acting before certainty becomes hindsight. That is security after signatures. That is security in the AI era.

Credit to: Daniel Levy, Threat Hunting Data Scientist

Edited by: Ryan Traill, Content Manager

References

[1] https://www.cyber.gov.au/business-government/secure-design/artificial-intelligence/careful-adoption-of-agentic-ai-services

[2]https://www.latimes.com/business/story/2026-02-26/hacker-used-anthropics-claude-ai-to-steal-mexican-government-data

About the author

Nathaniel Jones

VP, Security & AI Strategy, Field CISO

Blog

AI

July 10, 2026

AIインフラがアタックサーフェスの一部に

AIインフラとアタックサーフェスの進化

多くの組織が生成AIを実運用環境に導入するなかで、企業のクラウド環境内に新たなインフラのレイヤーが出現しています。それはAIゲートウェイです。AIゲートウェイはユーザー、アプリケーション、基盤モデルの間に位置し、多くの場合クラウドの特権アクセスを保持し、さまざまなAIサービスへのアクセスを大規模に管理しています。

AIゲートウェイとは？

AIゲートウェイはユーザー、アプリケーション、基盤モデルの間に位置し、多くの場合クラウドの特権アクセスを保持し、さまざまなAIサービスへのアクセスを大規模に管理しています。

こうした役割から、AIゲートウェイは企業のアタックサーフェスのますます重要な一部になりつつあります。AIゲートウェイが侵害されれば、攻撃者に対して計算リソースへのアクセスだけでなく、クラウドアイデンティティ、モデルサービス、機密性の高いプロンプト、そして他の接続されたシステムへのアクセスも提供してしまいます。

このブログでは、Amazon Bedrock サービスに接続されたAIゲートウェイが侵害され、その後暗号通貨マイニングインフラとの通信が観測された事例をダークトレースがどのように調査したかを解説します。問題のインスタンスは、その構成、ならびに関連するIAM（Identity and Access Management）ロールから、Amazon BedrockでホスティングされるAIサービスへのゲートウェイとして機能していることがわかりました。疑わしい侵害アクティビティが発生した後、このホストは既知の暗号通貨マイニングインフラに繰り返し通信を行い、その後シャットダウンされた様子が観測されました。Darktrace はこのアクティビティを検知し、Enhanced MonitoringおよびManaged Threat Detectionサービスを通じてエスカレーションを行いました。

この事例では最終的影響は不正な暗号通貨マイニングでしたが、このインシデントが注目に値するのはその発生場所です。侵害されたアセットは、クラウドインフラ、アイデンティティ、各種AIサービスの交差する場所に位置していました。最近の調査では、LiteLLM等のAIゲートウェイが、認証情報、モデルへのアクセス、クラウド権限を中央管理するその能力から、攻撃者にとって魅力的な標的となる可能性が明らかになっています。このアクティビティと公開されているLiteLLM脆弱性を直接結びつける証拠は見つかっていませんが、このインシデントは、AIインフラを個別のアプリケーション層として見るのではなく、重要なアタックサーフェスの一部として扱う必要性があることを表しています[1]。

暗号通貨マイニングがクラウド侵害後のアクティビティとしてよく見られる背景

暗号通貨マイニングはクラウド環境において、侵害後のアクティビティとして収益性の高いものとなり得ます。クラウド資産にアクセスできるようになった後、攻撃者はマイニングソフトウェアを展開して被害者の計算リソースを悪用し金銭的利益を得ることができます。この種のアクティビティは多くの場合機会主義的なものであり、露出したサービス、弱い認証情報、漏洩したアクセスキー、脆弱なアプリケーション、あるいはクラウドワークロードの設定ミスなどを標的として実行されます。

典型的なクラウド上での暗号通貨マイニング侵入には次のようなアクティビティが含まれます：

露出したあるいは脆弱なクラウドインフラの特定
露出したサービス、認証情報、またはアプリケーションの脆弱性を通じたアクセスの獲得
マイニングソフトウェアのダウンロードおよび実行
マイニングプールインフラへのアウトバウンド接続を繰り返し確立
アクティビティが検知され停止されるまで継続して計算リソースを消費

‍

この事例において注目すべき要素は暗号通貨マイニングだけではありません。それが発生した場所が、AI関連アクティビティをサポートするクラウドインフラ上だったことです。この事例は、AIサービスを実現するためのアセットも、よくあるクラウド侵害リスクにさらされる可能性があることを示しています。

Amazon Bedrockに接続されたAIゲートウェイの侵害を調査

2026年6月12日、DarktraceはLiteLLM-Proxyという名前のAmazon Web Service (AWS) EC2インスタンスから暗号通貨マイニング発生中とみられるアクティビティを観測しました。このインスタンスはLiteLLMアクティビティをサポートしており、Amazon Bedrockリソースへのアクセス権を有するインスタンスプロファイルと関連付けられていました。

AIゲートウェイは大規模言語モデルへのアクセスを中央管理するよう設計されており、多くの場合AIアプリケーションに対する認証、ルーティング、ログ、ポリシー適用を扱っています。セキュリティの視点から見ると、クラウド権限、モデルアクセス、アプリケーションワークフローを単一の制御ポイントに集約する役割も果たしています。その結果、AIゲートウェイの侵害は、侵害されたホストだけにとどまらない影響を及ぼす可能性があります。

確定的な初期アクセスベクトルは確認できませんでしたが、このアクティビティはインターネットに接続されているシステムの侵害でよく見られる次のような順序に従っていました。ブルートフォースアクセス、ペイロードの投下、そしてマイニングプールインフラに対する繰り返しのアウトバウンド接続です。

ステージ1: インターネットに露出したSSHからの初期アクセス

暗号通貨マイニングアクティビティが観測される前、LiteLLM-Proxy EC2インスタンスはSSH（ポート22）が0.0.0.0/0に対して開かれ、外部に公開されていました。

*図1：EC2インスタンスがSSHポート22に対してすべてのインバウンドトラフィックを許可している設定ミスをDarktraceが警告*

暗号通貨マイニングアクティビティに先立って、Darktraceはこのインスタンスに対する大量のインバウンド接続の試みが外部IPアドレス（主に145.241.123[.]102）からポート22に対して行われていることを観測しました。これはブルートフォースアクティビティを示唆するものです [2]。これらの接続の多くは短命であり、数秒しか続いておらず、スキャニングまたはログインの失敗を示していました。

*図2：Darktraceがデバイスのポート22に対する不審なインバウンド接続試行を検知*

入手できたテレメトリーではこれらのインバウンドSSH接続のいずれかが認証の成功につながったかどうかの確認に至らず、このアクティビティが初期アクセスベクトルであると断定することはできませんでした。しかしながら、SSHの露出、外部IPアドレスからのインバウンド接続、それに続くマイニングアクティビティは、SSHがアクセス経路の可能性が高いことを示唆しています。

ステージ2: AIゲートウェイへのXMRigマルウェアのダウンロード

最初に観測されたマイニングプールへの接続の後、このEC2インスタンスは3.42 MBのデータをポート80上のHTTP接続を介して外部エンドポイント185.62.1[.]8にダウンロードしました。このエンドポイントは暗号通貨マイニングマルウェアXMRigを含むZIPファイルをホスティングしていました[3][4]。ホストレベルのログは入手できなかったため、ダークトレースはマイニングツールがどのように実行されたか、あるいは前のSSHアクティビティがペイロード投下を直接的に可能にしたかどうかを確認できませんでした。しかしながら、ダウンロードのタイミングとその後ほどなくマイニングプールへの接続が繰り返されたことは、このインスタンスが侵害されて不正な計算アクティビティに使われたという評価を裏付けています。

ステージ3 – 侵害されたAIゲートウェイが暗号通貨マイニングインフラと通信

わずか数分後、DarktraceはLiteLLM-ProxyEC2インスタンスがHTTPs（ポート443）でホスト名pool.hasvault[.]proに対して接続していることを確認しました。最初の接続の後、同じホスト名に対して繰り返しアウトバウンド接続が観測されました。これは、侵害されたホストがマイニングインフラと通信しワークを受け取り、結果を送信するという、暗号通貨マイニングプールとの通信のパターンと一致しています。

このアクティビティがDarktraceのEnhanced Monitoringモデル“Compromise / HighPriority Crypto Currency Mining”をトリガーし、ダークトレースのSOCにより顧客に対してエスカレーションされました。また、このアクティビティはCyber AI Analystによって分析され、関連するイベントが1つの調査ナラティブにまとめられました。これにより、影響を受けたクラウドアセットからマニングプールへの繰り返しの接続を特定することができました。

*図3：CyberAI Analystによる暗号通貨マイニングアクティビティの調査*

ポート443上のHTTPSの使用にも注目すべきです。なぜならば、単独で見れば、このトラフィックそのものは疑わしく見えないかもしれないからです。しかしこのケースでは、接続先、接続の量、そして類似のアクティビティが他にないことなどが、この通信を疑わしいものとして特定するのに必要な、動作のコンテキストを提供することになりました。

ステージ4: Managed Threat Detectionサービスによるリソース乱用の特定

暗号通貨マイニングアクティビティがダークトレースのManaged Threat Detectionサービスにより検知され、ダークトレースのSOCによりレビューされました。レビューの結果、このアクティビティは顧客向けにエスカレーションされました。このエスカレーションにより、顧客はAWS環境で現在発生中のリソースの乱用について、タイムリーな通知を受けることができました。

ステージ5: クラウド認証情報の不正使用とみられる疑わしいIAMアクティビティ

これとは別に、6月13日、Darktraceは別のIAMユーザーから発生した疑わしいアクティビティを検知しました。

*図4: DarktraceのAdvanced Search機能が別のIAMユーザーが実行した疑わしいアクティビティをハイライト*

まず、このユーザーは “GetSendQuota”イベントを試行している様子が見られました。このアクションは少なくとも過去3か月間にこのアカウントによって実行されたことのないアクションです。また、このコマンドのソースIPアドレスは14.176.1[.]47でした。地理位置情報はベトナムであり、このユーザーのアクティビティがAmazon IPアドレスから最も多く見られた場所です。さらに、このアクティビティに対してAWS CLIが使用されており、これもこのユーザーにとって通常とは異なる振る舞いでした。このことは、Darktraceの“IaaS / Unusual Activity / UnusualAWS CLI Activity”モデルによって検知されました。

*図5: Darktraceによる “GetSendQuota” イベントの検知*

このIAMユーザーからは、長期アクセスキーを使った疑わしいアクティビティがさらに観測されました。中でも、“InvokeModel” および “ListFoundationModels”コマンドの失敗が検知されており、モデル列挙や起動などAmazon Bedrockサービスとのやり取りを試行したことがわかります。これは前日観測されたLiteLLM侵害への関連を思わせますが、2つのイベントを確定的に結びつける証拠は不十分でした。

“CreateUser”コマンドの試行も注目に値します。なぜなら要求されたユーザー名は意味が薄いものであり、新しいアカウントを作成することにより永続性を確立する試みと見られるからです。このアクティビティはDarktraceのモデル“IaaS / Admin / New AWS UserAccount Creation”をトリガーしました。

2つのインシデント間に結びつきは確認できなかったものの、このIAMアクティビティには重要な意味があります。これは、クラウド侵害の調査においてワークロードのテレメトリーとコントロールプレーンのテレメトリーの両方を取り入れることの重要性を表しています。EC2暗号通貨マイニングアクティビティが計算リソースの乱用を示す一方、IAMアクティビティは認証情報の侵害や長期アクセスキーの不正使用、そしてクラウトサービスの不正使用の可能性を示唆しているからです。

AIインフラ保護のための重要な教訓

このインシデントの重大性は暗号通貨マイニングアクティビティそのものではなく、それが発生した場所にあります。侵害されたシステムはAmazon Bedrockサービスへのアクセス権を持つAIゲートウェイとして機能し、クラウドインフラ、アイデンティティ、そしてさまざまなAIオペレーションの交差する場所に位置していました。組織がAI機能を実運用環境に導入していくなかで、これらのプラットフォームは、露出したサービス、認証情報窃取、クラウドの設定ミスなどを通じて攻撃者がすでに狙っているアタックサーフェスの一部となりつつあるのです。

このケースでは詳細な侵入経路は特定されておらず、ワークロードの侵害と調査中に検知された疑わしいIAMアクティビティの間に決定的なつながりは確認されませんでしたが、これらのイベントは全体的な現状を裏付けています。つまり、AIインフラは個別のテクノロジースタックとして扱うのではなく、クラウド環境全体の一部として保護しなければならないとうことです。

このケースでは、最も目立った侵害の兆候は暗号通貨マイニングインフラとの通信でした。しかしここで得られたより重要な教訓は、このインシデントの全貌が理解される前にDarktraceのビヘイビア分析により明らかになった、高い権限を持つAI関連アセットを取り巻くリスクです。AIゲートウェイによりクラウド権限、モデルアクセス、アプリケーションワークフローがますます集約されるなかで、防御者は個別のアラートに集中するよりも、ワークロード、アイデンティティ、サービスの間でどのように動作がつながっているかを理解することに重点を置く必要があるでしょう。

協力：Angel Arribas Lopez (Associate Principal Cyber Analyst)、Nathaniel Jones (Field CISO/VP Threat Research)、Emma Foulger (Global Threat Ops)、Mark Turner(Security Researcher)

編集：Ryan Traill (Content Manager)

付録

Darktraceによるモデル検知結果‍

· Compromise / High Priority Crypto Currency Mining

· Compromise / Monero Mining

· Device / Internet Facing Device with High Priority Alert

· IaaS / Unusual Activity / Unusual AWS CLI Activity

· IaaS / Admin / New AWS User Account Creation

MITRE ATT&CK マッピング‍

初期アクセス – 外部リモートサービス – T1133

初期アクセス – 有効なアカウント – T1078

実行 – コマンドおよびスクリプトインタプリタ – T1059

永続化 – アカウント作成 – T1136

探索 – クラウドサービス探索 – T1526

影響 – リソースハイジャッキング– T1496

参考資料

[1] https://docs.litellm.ai/blog/security-update-march-2026

[2] https://www.abuseipdb.com/check/145.241.123.102

[3] https://urlscan.io/search/#185.62.1.8

[4] https://www.virustotal.com/gui/file/85de36ff66fae9f4b059cbedf6d36e017ebc26c828f99f911a96e78636f21200/community

About the author

Angel Arribas Lopez

Associate Principal Cyber Analyst

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

デモを予約

Check out this article by Darktrace: Explore Internet-Facing System Vulnerabilities

Explore Internet-Facing System Vulnerabilities

Exchange server compromise

Confluence server compromise

GitLab server compromise

Log4j server compromise

Round-up

Footnotes

AIインフラがアタックサーフェスの一部に

Security After Signatures: Operating in a World of Pre‑CVE Disclosure Exploitation, Collapsed Trust Boundaries, and Autonomous Systems

ブログをお楽しみ頂けましたか？

More in this series

Security After Signatures: Operating in a World of Pre‑CVE Disclosure Exploitation, Collapsed Trust Boundaries, and Autonomous Systems

AIインフラがアタックサーフェスの一部に

Securing AI: Analysis of the Complete Security Stack with Governance and Controls

Blog

AI

July 13, 2026

Security After Signatures: Operating in a World of Pre‑CVE Disclosure Exploitation, Collapsed Trust Boundaries, and Autonomous Systems

Exploitation before disclosure

Identity is the real control plane

AI accelerates the threat

1. AI as an Attack Multiplier

2. AI as an Attack Surface

AI as a trusted but dangerous actor

Conclusion

References

Blog

AI

July 10, 2026

AIインフラがアタックサーフェスの一部に

AIインフラとアタックサーフェスの進化

AIゲートウェイとは？

暗号通貨マイニングがクラウド侵害後のアクティビティとしてよく見られる背景

Amazon Bedrockに接続されたAIゲートウェイの侵害を調査

ステージ1: インターネットに露出したSSHからの初期アクセス

ステージ2: AIゲートウェイへのXMRigマルウェアのダウンロード

ステージ3 – 侵害されたAIゲートウェイが暗号通貨マイニングインフラと通信

ステージ4: Managed Threat Detectionサービスによるリソース乱用の特定

ステージ5: クラウド認証情報の不正使用とみられる疑わしいIAMアクティビティ

AIインフラ保護のための重要な教訓

付録

Darktraceによるモデル検知結果‍

MITRE ATT&CK マッピング‍

参考資料