Threats

Aokigahara Malware Steals Login Credentials for AI Platforms

Key Takeaways A new Windows infostealer, Aokigahara, is actively compromising login credentials and session data from popular AI platforms. The malware targets local authentication files and...

Jennifer sherman
Jennifer sherman
October 27, 2026 5 Min Read
2 0

Key Takeaways

  • A new Windows infostealer, Aokigahara, is actively compromising login credentials and session data from popular AI platforms.
  • The malware targets local authentication files and application caches of AI tools like ChatGPT Desktop, Claude, and Cursor.
  • This threat does not indicate a breach of the AI services themselves but rather exploits locally stored user data.
  • Organizations must integrate AI application sessions into existing infostealer response plans, treating them as critical assets.

A novel Windows-based information stealer, dubbed Aokigahara, has emerged, specifically designed to pilfer login credentials and session tokens from a range of AI development platforms. This sophisticated malware targets prominent tools such as ChatGPT Desktop, Claude, Codex, and Cursor, representing an evolving threat landscape where cybercriminals are increasingly focusing on the burgeoning AI ecosystem.

Table Of Content

The Aokigahara malware operates by systematically scanning infected systems for authentication files and application storage areas. This method extends traditional password theft techniques to encompass the AI tools that developers frequently utilize in their daily operations. According to Researchers from Check Point identified Aokigahara in the wild and detailed its capabilities in an October 23 research report.

It is crucial to understand that this threat does not signify a direct breach of the AI services themselves. Instead, attackers are exploiting locally stored information on user machines to gain unauthorized access to accounts. The increasing prevalence of similar AI coding agent theft underscores the necessity of safeguarding tokens, saved connections, and project records with the same rigor applied to browser passwords and cloud keys.

Deep Dive into Aokigahara’s Capabilities

Check Point’s research suggests that Aokigahara might be an evolution of earlier malware variants known as Kamidere or Stygian. While the precise monetization strategies employed by the operators of this infostealer remain unconfirmed, its technical capabilities are extensive.

A dedicated module within Aokigahara, named aisessions, is specifically engineered to harvest data pertinent to a wide array of AI-related applications. This includes Claude Code, Claude Desktop, Codex, ChatGPT Desktop, and Cursor. The module’s reach extends further to encompass Windsurf, GitHub Copilot, Gemini CLI, JetBrains AI Assistant, Amazon Q, and Kiro. Beyond AI tools, the malware also queries the Windows Credential Manager for stored login information.

The collection process is granular, targeting specific files and folders for each AI platform. For instance, Claude’s credential files and desktop application directories are primary targets. Codex sees its authentication, configuration, and environment files compromised. Cursor’s local storage databases are also within scope, while ChatGPT Desktop’s application cache is thoroughly searched. Although these locations may contain valuable secrets, it’s important to note that not every targeted file will necessarily yield a password or session token.

Previous analyses of credential-stealing malware have highlighted a growing interest in AI application data. The actual risk posed by Aokigahara hinges on the validity and reusability of the exfiltrated secrets. Functional session tokens, for example, could allow attackers to bypass password authentication entirely, granting immediate access to victim accounts.

Such unauthorized access could expose sensitive private conversations, uploaded proprietary materials, or critical development information. Attackers might also leverage compromised accounts to consume paid services or resell access on underground markets. Check Point outlined these as potential consequences, emphasizing that they represent possible outcomes rather than confirmed activities by Aokigahara’s operators. The firm also raised the specter of data extortion, where the exposure of sensitive business content could be used for blackmail.

Beyond its AI-centric focus, Aokigahara incorporates a separate ide module designed to broaden its data collection. This module targets Git credentials, SSH files that may contain private keys, and critical configuration data for AWS, Azure, and Google Cloud environments. In addition to these, Aokigahara is capable of stealing browser credentials and cookies, capturing screenshots and clipboard contents, and gathering data from wallet, messaging, VPN, and email applications.

Evasion, Encrypted Uploads, and Incident Response

Aokigahara demonstrates a sophisticated approach to evasion, integrating optional checks to hinder malware analysis. Depending on its configuration, the infostealer can detect debugger indicators, virtual machine environments, hardware resource limitations, process counts, and even mouse movement. Furthermore, it can cease operation if keyboard or Windows interface language settings match an excluded list, which notably includes Russian, Ukrainian, and several other regional languages.

The malware’s configuration is secured using AES-256-CBC encryption. This configuration dictates which of its built-in collection modules are activated during execution, stores server details, manages communication settings, and defines fallback destinations. A critical observation by researchers is that disabled collection modules remain embedded within the executable, implying that a sample’s full capabilities may exceed those enabled in any given run.

For data exfiltration, Aokigahara packages collected information into ZIP archives, which are then encrypted using AES-256-GCM. An X25519 key exchange protocol facilitates the establishment of the encryption key. The upload process involves distinct requests for opening an upload session, transferring each category of collected data, and signaling completion. Analysis of the latest samples reveals the use of a User-Agent string resembling a legitimate Windows application, though this alone is insufficient to confirm an infection.

Upload headers play a crucial role in categorizing the exfiltrated data, distinguishing browser data from AI session information and other stolen materials. While a final completion message indicates the end of the upload phase, it does not independently verify that every stolen archive successfully reached the command-and-control server.

The emergence of Aokigahara aligns with a broader trend, exemplified by incidents involving counterfeit Gemini npm packages that targeted local data from various AI tools. This parallel campaign, while not directly linked to Aokigahara’s delivery methods, highlights the growing attractiveness of developer workflows as points of compromise for reusable access credentials.

For cybersecurity defenders, the collection patterns observed in Aokigahara necessitate vigilance. Monitoring for unexpected processes accessing AI authentication stores, editor databases, cloud credentials, and SSH materials is paramount. Network-level findings should always be corroborated with endpoint evidence rather than being treated as standalone proof of compromise. Check Point has also released a YARA rule to aid in the detection of this malware.

What You Should Do

  • Isolate Infected Devices: Immediately disconnect any suspected infected device from the network to prevent further compromise.
  • Revoke AI Sessions: From a clean system, revoke all active AI platform sessions associated with the compromised user.
  • Rotate Exposed Keys: Change all cloud keys, API tokens, and SSH keys that may have been exposed.
  • Review Connected Accounts: Scrutinize all connected cloud and repository accounts for any unauthorized activity.
  • Reset Passwords: While crucial, understand that resetting passwords alone may not be sufficient if session tokens were stolen.
  • Audit AI Account Activity: Review AI account logs for unusual conversations, uploaded files, or service consumption.
  • Integrate AI Sessions into Incident Response: Treat AI application sessions as critical assets within your existing infostealer response plans.
  • Deploy Detection Rules: Implement Check Point’s published YARA rule for Aokigahara to enhance detection capabilities.

The findings related to Aokigahara concern theft from compromised endpoints, not a demonstrated vulnerability within ChatGPT, Claude, Codex, or Cursor themselves. Organizations must therefore proactively integrate AI application sessions into their established infostealer response frameworks, recognizing their elevated importance beyond ordinary application settings.

Indicators of compromise (IoCs):-

Type Indicator Description
IP address and port 2[.]26[.]252[.]48:8080 Aokigahara command-and-control infrastructure
Domain dfwioeiofwr[.]info Aokigahara command-and-control infrastructure
Domain dobeqnesmot[.]cc Aokigahara command-and-control infrastructure
Subdomain gate[.]dfwioeiofwr[.]info Aokigahara command-and-control infrastructure
Domain hjazfioe[.]surf Aokigahara command-and-control infrastructure
Domain kkeyyou[.]cc Aokigahara command-and-control infrastructure
Domain neihite[.]cc Aokigahara command-and-control infrastructure
Domain somelitersofblood[.]pw Aokigahara command-and-control infrastructure
SHA-1 8d6c86e0af7807f9b5baf15e9f076bb81e012515 Dropped clipper component

The following table reproduces all 150 Aokigahara sample SHA-1 hashes. Each cell represents a separate sample; columns do not indicate relationships between samples.

Aokigahara SHA-1 Aokigahara SHA-1 Aokigahara SHA-1
007137599bc4d69ee952a5228b5fcb2c3b9a95a7 00d52c7e759d878f0d8a117502f6d854af13f007 04eddccb79db41e03a4ee8542e8a33d7f1d9fe18
057a2d1dedeb93b7765c455a9e0c469e7cbe4843 079ddf5cc7d6ca2e02f860f87d4d18c83db9d331 0b74ce90a39278fb97fe94f6306af71cbe679503
0c46ac5555cf8f305c1345a21d0802c52bf52cd8 0d8d0bad2e08856ba379e425e2e90e49506af631 0e76150cee3f2b89f0fbe39106ab2898db109f09
0f69792cb926a416766d47a30a05009c13be0d64 130f2710cf9863bdd6724e8912600505b6fb7ef2 133dc00ad81638e6b0a2f8f163b4d694ce477165
151d8013ca3ab3dab9cd37f4ed9870cc83e87a77 1701db4fa798b309cb25227408406dde4c69de8a 1835f1e0b1eda3dc4b39682ecad3eba88cbbc44e
18e041b5a0c668392ff4d036f3f9660c2698311e 1a6a6e08a01cc3f1455af2c4ecd59a3d7802cd7a 1a7248720f3ac79082841b006582f20799b314a9
1a97a18db74f2b6df984c9a188a79b4e65a06b81 1cb6ccfdcb277a0277729443d15f17a441fa0eda 1cb8dc11bdb8baf342b83079fb650cd6bb1b132d
1e072236f87cd914dc583ca2163460d656bfcfba 1e94d7f1d501724076fcf98459e14b68321d45f7 2045f165d737f2d43ba603d444dfa5d9e6f05805
211e45f306b84a82cfe736635c176a424e7479dc 21ef070e39f651118a804882374dc2dfa3e38666 2362a326070d83c167251873b389bd783781cf6c
241e37b8fc5ffeefbd1d89a82b74cf2d92b49eac 244371fc1642f0548b807b0a4a2e7ea5d7b4c82a 28099fc5a8cbbd40173f8ac0aded9044fdb4a598
2cdc77108c585d14b1567053b8f8a58ea78c255f 2f30b116ca3dc48ff5783c3bb7391d671866e27b 30737fa46f2ff93e33f45008cf306233954ea898
324e0fffc7977f8ff591993819123d33c2970953 3586eeaf3ffe2c3a4f1d005aee863b7aee7cc7bd 3696baa239c479f9f5103618214972890f1a473b
385e21a96687a33f79f9cab0627aec7218e1a8f4 38b2c29b46086b417181deb958e8b276a9d1fbf4 3aaeb49864108c92af50feeb5d9e3b3efaa39818
3c56a0c3c59a1425bfdcdffda9e8b308f478b147 426bea548a24f7f39c279770b5dd21fbed7f7328 4417d2645adb6e7e94959daf483f93058500840f
4485feb1385975ef13f08148cdc679f50fa0cc6d 456414c0acf463e06f95f5bd3d8355758fca6b53 4595d1875441ffb8c076856e366d1f4f5c74a679
48c6b66dc4b4d40d72b7460e6fd670b6c03d5d40 4b3a4ad0acd43fc8796147381074a98e8e576904 4c07021f8b0cdd3ca334a13f9af2bade9e699dc2
50060c0fc5fa8932671eb24b03b05d00c2e99951 51ae6cad54f33846f83a352457b9a44ef8b7a61d 59bbaca8a848215e7e4af7cec320432c51b0b812
5ef7eb755f1e3f6bf8586794777466a639057dcb 5fadf206dc163f63ebf66f984042b00c67121813 5fe11fbb89d33efa712ffb8590626bd88fecab8f
61a024f5a42a9a91b0a9871a1613c7fb3ef5613b 62dcc9ba2a5eee9dd941ad08766603c89a45869b 6393c0d3fb76f771a9418c2d152c621929ee002b
63cc5033532cd00bda3d4968b1257629e613505a 682dd6392004e1131b1aa3f2351a0bc7adb84d14 699c809a116106da40b3f641c11b0db34d0f1c5e
6a09fffca381f6624d2def1083284ca486306e7c 6c860007f8f49de8e9848b538bd7acdc6bfcdf55 6e073a196a9bfa01da418b5783c2395fb22c7ec9
6eb8f17dba69f6d25b7f5c529b25d199707b69e2 6f7d5935602c188b1e3066b4ab9aeef7b85a8ec2 70915fac07af4ad7acc1868fe05dc00b98206f0f
70d4007b11e56752fffb81f71c90987f4bb441fa 70e91644802053cfa68be6ec0f94207282ff27e4 7233a15327dc2f0cca0292e6352eebf6e6e090d8
72aa6417a19c45dabdd6c79087fb13c73ae7b823 7cc6c75fad93b2482a9c5f4522530007c72f99e0 82ebfc09c98bd211ff8b08efcd8ec8b9ce1fb4be
83b43ce0ee9881201451c0604fb5b792bd4581f5 86230c07e638b85bc8010df1aa9e33ca61fda4cc 89f296aef7988bdd59109e4263d0a31c886770aa
8b0cf62382f620bbbf5d176295d63ed522ab9c2c 8c3111a61fbf30e86b8206397ce3c0017e5ea509 8c4c1d8325adb463c34694ce3a5a851e97b81a6f
8d7a81119ce7be6eb452192f08fa2669228ff597 8d860ca64ce0cbf40871fd66b9e3a513e73cda90 8dc40b7208b8c0d95d1ab8dda9a56fff790b3ddd
8e9fd1274b77ff9726c4637d008a9d599c9d5f1e 9003310bf73dc337b8b50131f1c0724a79161aac 90c057dafefbee423e3014266218272d454a0aa3
912ed8ce7767f94f62de6a30fb8e9df7c0c512f7 9369e0e5eae22cff71b375ad285ea89471f25101 97c80e2d382e6e55f61702da3afb838fdec2f4f1
98c6649cc256dfadf1af939da46f037b05c0f2e5 99b28a20851a3ed29769a469f4dd31ebbb8a83de 9d424ce23753ffddc7b39477222123b0f2c33066
9f05045f5447d79c8c0d0f5cb589fd85c89620d3 9f20034441e78574aec7ad39842a342f939eaf6a a00da33342a5ea5168d851a5c93109e07cf95c49
a2abdf1af38da9995a3b6672ce6c34e9e17c39a4 a4871f4a61faf74e1cceaffe26901f6076c9f37e a63ded811d412cd0f859891b0c1f066b87bb1b80
a84148b4d408544325c20ee04299ae95c762b6a2 ad960d2abb1999a7260c5f498532a5bb8754c2a5 ae9d6b1a32a45b05edb920538e538a3d6a5fe439
af196456d2ebf9556ab239661922565e1bca9ac0 b335483800d35a29c25767c917725e24c21b0f2b b33e59c598845d2547ddc7f654115df2284bb3ae
b36d0796e33f808eaeb33b33908327a410680c56 b4491f1919a88deedebe7fba56b593e7d6655458 b523d56d82485253f2e3619437c8398b7ee0f20d
b607025d3e06d1e8cffcd398c64871d7fcd92bf4 b6f0dfd3aa2fc7b8b2ad432c4a4276da23ceb46b bab4c910e3c98085bf228519f346d071a75e44e3
bad339c5ad65808285b0e87498f2ea6f669417d9 c00aceaa40b38e48f662ac927bb86302934aaf96 c2e28ba5523c1f3a545a0b4a00a9d5e9c8ae7583
c323c59815082ac842f0bafcc01f537444e1cc5a c5b50b5e14dfb5e028dfd88a4e4fb960dd650b52 c810acdf30bcdfc6e8021e7f62a3bcce0828c2ee
c8456bb71a74ed6305455c318295b8a39e9277f3 cbdff0df0dbdbc3c1845d208661dd75fd335ef93 cc12fcb9e5e6f7afde66f5c3452ed6e3cd6d5390
cc708ae3d7aa084279122c0db7250096f750d6b0 d34491da5aa159314129c0ff67f957d873703e35 d3c90275dd7279a791d717676be15789376f7aa7
d8bfff5139b54b6f98f07c051cd25d7aa465fec8 da986a3cb2ce67456ef374fbdd957d370a1a4367 daef439e1f110d624ce50fe0f9d9d24f61f7497b
db3332ce75676389b79cf3603d81cb231cb7c19c db69c7c930fd5dc5097847979a965fdea4171d5b dba5fef27b53ff479a4d7046cd1f22710081a16c
de9d152010819c8fb5cb611a4de920399e69cfeb dfd1540c0185e74ec65acfcae1af7ba0da36c774 e052ff49190f54e0ffc52f7b3f7b5e7b9e121c4c
e3af6d9e3b35e5e04feda3cde96256dd3b099624 e4d93fa1b2d0fd26e8b877511ea33badb4628169 e5c43abd1c212b2f2a61d5e1cb7821406ca5dc8e
e8d09eb2e5857fde39b44aeb48078f5396809026 ec7cd90d56e5b73caa9d4eb0472bfee68d716121 eec7ece53d47e50940584f4dec684a24a78f42f2
f0b23f420659babb3a730b122259440091c53c12 f30e9730db0663f87c67a4b9a0cf12ee59b1621a f35e6d2936c0f6dc26bdc68f6d1599cf438a863a
f4184e1bb61e19d190555c50fdb9d4591767bf53 f47d45eea1c42b066f8537baf72ffa266646bf62 f5993e64a18e7494b25a185de9f63a1df961b99a
f633fc9fd16dadca2725d3d54aaf13ea1a62174f f6d950a5a8e5ba336f748b66745c6c1f189d3362 f8cf33b18e9932d285696fee40a7fee2a87341c2
f8dbc552c89bef758e4fcbd9942d06bc0c11db0a f990bf4f05dfa4d86b3e3bc0a54f62e91b83f2c6 fa61cf3bebefd88ea9d06007a4b85791aa851c8a
fbe8708247fabd8

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.

Tags:

AttackBreachMalwareThreatVulnerability

Share Article

Jennifer sherman

Jennifer sherman

Jennifer is a cybersecurity news reporter covering data breaches, ransomware campaigns, and dark web markets. With a background in incident response, Jennifer provides unique insights into how organizations respond to cyber attacks and the evolving tactics of threat actors. Her reporting has covered major breaches affecting millions of users and has helped organizations understand emerging threats. Jennifer combines technical knowledge with investigative journalism to deliver in-depth coverage of cybersecurity incidents.

No Comment! Be the first one.

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts