Aokigahara Stealer Grabs Your Claude, Cursor and Copilot Sessions

A new infostealer called Aokigahara goes after the usual browser passwords and crypto wallets, but it also carries a dedicated module for the logins of AI assistants and coding tools, Check Point Research says. The list includes Claude Code, Codex, ChatGPT Desktop, Cursor, Windsurf, GitHub Copilot and Gemini CLI. Check Point says the malware is still under active development and may be a successor to a stealer earlier known as Kamidere or Stygian.

2.26.252[.]48:8080
dfwioeiofwr[.]info
gate.dfwioeiofwr[.]info
dobeqnesmot[.]cc
hjazfioe[.]surf
kkeyyou[.]cc
neihite[.]cc
somelitersofblood[.]pw

Source article: Infostealer riding the AI Train
Publisher: Check Point Research (YARA rule credited to @Tera0017/Check Point Research)
Published: Oct. 23, 2026
URL: https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/

Executive Summary

Aokigahara is a newly observed Windows infostealer that Check Point Research says may be a successor to a stealer earlier known as Kamidere or Stygian (Check Point’s YARA rule still matches the string “kamidere”). Check Point describes it as under active development and does not say how it is delivered, whether it persists, or who operates it, so this is a capability profile rather than a campaign with known victims. On execution it can exit if the keyboard layout or Windows UI language belongs to a CIS-region language, or if it detects a debugger, virtual machine or sandbox; it then decrypts an AES-256-CBC configuration from a .stgcfg PE section and runs whichever collection modules a bitmask enables. Those modules take saved logins and cookies from about 30 Chromium browsers and Firefox, desktop and browser-extension crypto wallets, Discord, Telegram, Steam and gaming accounts, Outlook and Thunderbird data, VPN and FTP configs, password-manager files, and documents from the Desktop, Documents and Downloads folders. Two modules set it apart: aisessions, which copies credential and session files for Claude Code and Claude Desktop, Codex, ChatGPT Desktop, Cursor, Windsurf, Kiro, GitHub Copilot, Gemini CLI, Amazon Q, JetBrains AI Assistant and AWS CLI/SSO, and queries Windows Credential Manager; and ide, which takes Git credentials, .ssh private keys and AWS, Azure and Google Cloud configuration. A clipper module drops a separate clipboard-hijacking component under C:\Microsoft\Vault. Stolen data leaves as AES-256-GCM-encrypted ZIP chunks posted over HTTP to /gate/handshake, /gate/chunk and /gate/complete, using the user agent Microsoft-CryptoAPI/10.0 in the latest samples. The objective is credential and session theft for resale or account abuse; on a developer workstation the impact reaches source repositories, cloud accounts and paid AI subscriptions, and a password reset alone does not cut off a stolen session. Hunt leads, false-positive expectations, Check Point’s published YARA rule and the indicators are detailed below.

ATT&CK chain: Defense Evasion → Discovery → Credential Access → Collection → Command and Control → Exfiltration

Hunt Lead Grid

Aokigahara C2 Infrastructure Sweep

Check Point lists the stealer’s command-and-control infrastructure as 2.26.252[.]48:8080 and the domains dfwioeiofwr[.]info, gate.dfwioeiofwr[.]info, dobeqnesmot[.]cc, hjazfioe[.]surf, kkeyyou[.]cc, neihite[.]cc and somelitersofblood[.]pw. The configuration holds a C&C base URL plus up to three fallback URLs, so one infected host may contact more than one of these. The detection logic is a retrospective search of DNS resolver logs, web proxy logs and firewall or NetFlow records for any resolution of or connection to these domains, and for any connection to 2.26.252[.]48 on TCP 8080, followed by a standing block and alert on all eight.

Use DNS query logs from every resolver (including endpoint DNS telemetry from EDR, which catches hosts that bypass the corporate resolver), proxy logs, and perimeter firewall or NetFlow data. Sweep back to at least September 2026; Check Point published on Oct. 23 and the samples span several versions, so 90 days is a sensible floor. Scope to all Windows endpoints, then prioritize developer, build and administrator machines, because those hold the AI-tool, Git, SSH and cloud secrets that make this stealer distinctive. Exclude only your own threat-intelligence and sandbox hosts. Blocking the domains is worthwhile, but infrastructure rotates quickly in active development, so pair this with the behavioral leads below rather than relying on it. Any hit from a user workstation means data has probably already left: escalate to incident response, isolate the host, and start credential and session revocation for every account and tool listed in the executive summary rather than waiting to confirm what was taken.

Hunt Lead Level: 1 (Junior Analyst)

ATT&CK: Command and Control → T1071.001 (Application Layer Protocol: Web Protocols)

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Check Point YARA and SHA-1 Retro-Hunt

Check Point published a YARA rule named aokigahara and about 150 SHA-1 hashes of stealer samples, plus one SHA-1 for the clipper component. The rule matches a PE file containing at least three of its eleven strings and two byte patterns, among them the HTTP header formats X-Chunk-Index: %u and X-Build-ID: %s, the result marker AKGRES, the AISessions/ archive folder, the named pipe format \\.\pipe\akg_%lx_%lx, the module-name sequence sysinfo/screenshot/clipboard, the family names aokigahara and kamidere, the /chunk path, a kmd_w_%d_%d.db temp-file format and the Roblox cookie name .ROBLOSECURITY. The detection logic is to run the rule across file stores, email attachment archives, sandbox detonation output and EDR file-capture repositories, and to search EDR and antivirus file telemetry for the SHA-1 values in the appendix.

Use whatever runs YARA at scale in your environment (EDR live-response or retro-hunt features, a malware repository, a mail-gateway attachment store, or a scheduled scan of user Downloads and temp folders), plus EDR file-hash telemetry. Sweep 90 days of stored files and hash telemetry. Because the stealer is under active development, expect new builds to miss the hash list sooner than the YARA rule; treat the hashes as a quick retro check and the rule as the longer-lived control. Hash blocking alone is the wrong primary control for the same reason. A YARA hit on a file that was executed is an incident, not a tuning case: isolate the host and begin session and key revocation. A hit only in an attachment store or a downloads folder with no execution evidence should still trigger a check of process history on that host.

rule aokigahara
{
    meta:
        author = "@Tera0017/Check Point Research"
        description = "Aokigahara Stealer"
    strings:
        $str1 = "X-Chunk-Index: %u" ascii fullword
        $str2 = "X-Build-ID: %s" ascii fullword
        $str3 = "AKGRES" ascii fullword
        $str4 = "AISessions/" ascii
        $str5 = "\\\\.\\pipe\\akg_%lx_%lx" ascii fullword
        $str6 = "sysinfo\x00screenshot\x00clipboard"
        $str7 = "aokigahara" ascii fullword
        $str8 = "kamidere" ascii fullword
        $str9 = "/chunk" ascii fullword
        $str10 = "%s\\kmd_w_%d_%d.db" ascii fullword
        $str11 = ".ROBLOSECURITY" ascii fullword
        $code1 = {45 69 ED B1 79 37 9E 48 C1 EF 11 48 C1 EA 20 41 31 D6 41 31 C6 41 31 FD 45 89 77 04 45 89 6F 08}
        $code2 = {41 4B 47 52 45 53 ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? EF BE AD DE }
    condition:
        uint16(0) == 0x5A4D and 3 of them
}

Rule reproduced from the source article’s “Yara Rule” section.

Hunt Lead Level: 1 (Junior Analyst)

ATT&CK: Credential Access → T1555.003 (Credentials from Password Stores: Credentials from Web Browsers) (our mapping to the stealer’s core behavior; the rule detects the binary itself, and Check Point publishes no ATT&CK mapping)

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Clipper Drop Under C:\Microsoft\Vault

The stealer’s clipper/clipper_drop handler writes a separate clipboard-manipulation component, built from embedded data, beneath C:\Microsoft\Vault; Check Point notes the handler itself does not touch the clipboard and that the wallets or address formats the clipper targets are unconfirmed. Check Point gives the clipper’s SHA-1 as 8d6c86e0af7807f9b5baf15e9f076bb81e012515. The detection logic is a file-creation and process-start rule for any file written to, or executed from, C:\Microsoft\ and in particular C:\Microsoft\Vault\, plus a hash match on the clipper SHA-1.

Use EDR file-creation and process-creation events, and run a one-time filesystem check for the existence of C:\Microsoft\Vault across the Windows estate, sweeping 90 days of telemetry. Legitimate Microsoft components live under C:\Program Files, C:\ProgramData\Microsoft and C:\Windows, not at the root of the system drive, so a root-level C:\Microsoft folder is itself worth a look; allowlist only by a verified Microsoft code-signing certificate, never by folder name. Scope to all endpoints, with priority on finance staff and anyone who handles cryptocurrency, since a clipper swaps copied wallet addresses at payment time. A hit should be escalated, and the user should be told to verify any recent crypto transfers against the intended addresses, because the clipper can keep stealing after the stealer’s one-time collection has finished.

Hunt Lead Level: 1 (Junior Analyst)

ATT&CK: not mapped (insufficient detail). Check Point confirms only that a component is dropped for clipboard manipulation; it does not describe how the clipper runs or persists, and T1115 (Clipboard Data) covers collecting clipboard contents, not replacing them.

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Aokigahara /gate Chunked Upload Pattern

Check Point documents the stealer’s upload protocol in detail. It first POSTs its 32-byte X25519 public key to /gate/handshake and only proceeds after an HTTP 200; it then POSTs each module’s output as an AES-256-GCM-encrypted ZIP to /gate/chunk with the headers X-Chunk (the module label), X-Chunk-Index and X-Total; and it ends with a POST to /gate/complete whose eight-byte body is COMPLETE. Every request carries X-Build-ID, X-Gate-Token, X-Session, X-HWID and X-KEX: x25519, with Content-Type: application/octet-stream and, in the latest samples, User-Agent: Microsoft-CryptoAPI/10.0. The detection logic is a proxy or network-sensor rule for HTTP POST requests whose URI path starts with /gate/ and ends in handshake, chunk or complete, or any request carrying the X-Build-ID or X-Chunk-Index header, raised in severity when the user agent is Microsoft-CryptoAPI/10.0.

Use web proxy logs that record method, URI path, user agent and, where available, request headers; network sensors such as Zeek or Suricata that log HTTP headers; and TLS inspection where policy allows, since header visibility needs decrypted traffic (the 8080 C2 above was plain HTTP). Sweep 90 days and run continuously. Do not alert on the user agent alone: Windows’ CryptoAPI legitimately sends Microsoft-CryptoAPI/10.0 on GET requests for certificate revocation lists and intermediate certificates, so the anchor is the POST method plus the stealer’s paths and custom headers, which CryptoAPI never uses. Check Point notes the user agent has changed across versions, which is another reason to anchor on the path sequence and headers. Blocking the CryptoAPI user agent would be the wrong control because it would break certificate validation. Any match is an escalation: the handshake-chunk-complete sequence means a full exfiltration ran. The rule generalizes to other stealers that reuse this gate-style panel layout.

Hunt Lead Level: 2 (Pro)

ATT&CK: Exfiltration → T1041 (Exfiltration Over C2 Channel); Command and Control → T1071.001 (Application Layer Protocol: Web Protocols)

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Unexpected Process Reading AI-Tool and Cloud Credential Files

The aisessions and ide modules read a fixed set of files under the user profile, including .claude\credentials.json and .claude.json (Claude Code), the Claude and ChatGPT Desktop package folders, .codex\auth.json, .codex\config.toml and .codex\.env (Codex), Cursor\User\globalStorage\state.vscdb and storage.json, Windsurf\User\globalStorage\state.vscdb, Code\User\globalStorage\state.vscdb (VS Code), .copilot\hosts.json, .gemini\.env, gemini-cli-workspace-token.json, .codeium\config.json, amazon-q\data.sqlite3, .aws\credentials, .aws\config, .aws\sso\cache (including kiro-auth-token.json), and JetBrains AIAssistantQuotaManager2.xml, plus Git credential files, .ssh contents and Azure and Google Cloud configuration; it also queries Windows Credential Manager. The detection logic is an endpoint rule that alerts when a process other than the owning application opens one of these files for read, with highest severity when the reading process runs from a user-writable path (Downloads, %TEMP%, %APPDATA%) or is unsigned, and when that process also enumerates Credential Manager.

Use EDR file-access telemetry (or Windows object-access auditing with SACLs on the specific files, if your EDR does not record reads) and process-creation events. Build the allowlist from the owning tools’ signed images (Claude, Codex, Cursor, Windsurf, Code, the AWS CLI, Git, OpenSSH and the JetBrains IDEs), plus backup, sync and security agents, matched by signer and install path rather than process name, because the stealer can name itself anything. Sweep 90 days and scope to developer, build and administrator machines first, where these files exist and hold the most value. Expect medium noise from IDE extensions and scripts that read .aws\credentials; tune by parent process and signer rather than by suppressing the file. A hit from an unsigned, user-path process is an incident: assume every listed token is stolen, revoke and reissue AI-tool sessions and API keys, rotate cloud access keys and SSH keys, and sign the user out of Git hosting. The rule generalizes to any stealer that targets developer secrets.

Hunt Lead Level: 2 (Pro)

ATT&CK: Credential Access → T1552.001 (Unsecured Credentials: Credentials In Files); Credential Access → T1555.004 (Credentials from Password Stores: Windows Credential Manager)

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Credential-Store Burst Followed by Chunked Upload

Aokigahara runs its enabled modules back to back: it reads browser Login Data and cookie stores across many Chromium profiles and Firefox logins.json, wallet files and extension storage, Discord LevelDB and Telegram tdata, the AI-tool and cloud files above, and files from Desktop, Documents and Downloads, then packages each module’s output as a ZIP and uploads the chunks in sequence. The detection logic is a stateful correlation per process: count distinct credential-store categories (browser, wallet, messenger, AI tool, cloud or SSH, password manager) opened by one process within a short window, such as five minutes, and alert when three or more categories are touched by a process that is not a signed, allowlisted agent and that process, or a child of it, then makes outbound HTTP POST connections within the next ten minutes.

This needs EDR file-access and network-connection events joined on process ID and host in one SIEM or data lake, with a maintained category map of credential-store paths (the paths in this report plus standard browser profile locations). Back-test 90 days and run continuously. Maintain an allowlist of backup, endpoint-security and migration tools by signer and install path, and baseline normal behavior on developer machines first, since IDEs touch several stores during startup. The correlation catches renamed or rebuilt samples that the hash list and infrastructure sweeps miss, and it fires even if the user agent or /gate/ paths change. Any hit should be escalated immediately to incident response with a full credential and session revocation plan, and the host should be reimaged rather than cleaned, because Check Point has not ruled out persistence. The pattern generalizes to most commodity infostealers.

Hunt Lead Level: 3 (Elite)

ATT&CK: Credential Access → T1555.003 (Credentials from Password Stores: Credentials from Web Browsers); Credential Access → T1552.001 (Unsecured Credentials: Credentials In Files); Exfiltration → T1041 (Exfiltration Over C2 Channel)

Sources: “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article)

Hash Appendix

Aokigahara stealer samples

SHA-1 values come from the source article, which lists them without file names, dates or build IDs. The source publishes no MD5 or SHA-256; with about 150 samples, they were not pivoted one by one through AlienVault OTX or VirusTotal for this report, so look up the SHA-1 there if you need the other formats for a specific sample.

MD5s

(not available)  # all Aokigahara samples below: source publishes SHA-1 only

SHA1s

007137599bc4d69ee952a5228b5fcb2c3b9a95a7  # Aokigahara stealer sample (file name not published)
00d52c7e759d878f0d8a117502f6d854af13f007  # Aokigahara stealer sample (file name not published)
04eddccb79db41e03a4ee8542e8a33d7f1d9fe18  # Aokigahara stealer sample (file name not published)
057a2d1dedeb93b7765c455a9e0c469e7cbe4843  # Aokigahara stealer sample (file name not published)
079ddf5cc7d6ca2e02f860f87d4d18c83db9d331  # Aokigahara stealer sample (file name not published)
0b74ce90a39278fb97fe94f6306af71cbe679503  # Aokigahara stealer sample (file name not published)
0c46ac5555cf8f305c1345a21d0802c52bf52cd8  # Aokigahara stealer sample (file name not published)
0d8d0bad2e08856ba379e425e2e90e49506af631  # Aokigahara stealer sample (file name not published)
0e76150cee3f2b89f0fbe39106ab2898db109f09  # Aokigahara stealer sample (file name not published)
0f69792cb926a416766d47a30a05009c13be0d64  # Aokigahara stealer sample (file name not published)
130f2710cf9863bdd6724e8912600505b6fb7ef2  # Aokigahara stealer sample (file name not published)
133dc00ad81638e6b0a2f8f163b4d694ce477165  # Aokigahara stealer sample (file name not published)
151d8013ca3ab3dab9cd37f4ed9870cc83e87a77  # Aokigahara stealer sample (file name not published)
1701db4fa798b309cb25227408406dde4c69de8a  # Aokigahara stealer sample (file name not published)
1835f1e0b1eda3dc4b39682ecad3eba88cbbc44e  # Aokigahara stealer sample (file name not published)
18e041b5a0c668392ff4d036f3f9660c2698311e  # Aokigahara stealer sample (file name not published)
1a6a6e08a01cc3f1455af2c4ecd59a3d7802cd7a  # Aokigahara stealer sample (file name not published)
1a7248720f3ac79082841b006582f20799b314a9  # Aokigahara stealer sample (file name not published)
1a97a18db74f2b6df984c9a188a79b4e65a06b81  # Aokigahara stealer sample (file name not published)
1cb6ccfdcb277a0277729443d15f17a441fa0eda  # Aokigahara stealer sample (file name not published)
1cb8dc11bdb8baf342b83079fb650cd6bb1b132d  # Aokigahara stealer sample (file name not published)
1e072236f87cd914dc583ca2163460d656bfcfba  # Aokigahara stealer sample (file name not published)
1e94d7f1d501724076fcf98459e14b68321d45f7  # Aokigahara stealer sample (file name not published)
2045f165d737f2d43ba603d444dfa5d9e6f05805  # Aokigahara stealer sample (file name not published)
211e45f306b84a82cfe736635c176a424e7479dc  # Aokigahara stealer sample (file name not published)
21ef070e39f651118a804882374dc2dfa3e38666  # Aokigahara stealer sample (file name not published)
2362a326070d83c167251873b389bd783781cf6c  # Aokigahara stealer sample (file name not published)
241e37b8fc5ffeefbd1d89a82b74cf2d92b49eac  # Aokigahara stealer sample (file name not published)
244371fc1642f0548b807b0a4a2e7ea5d7b4c82a  # Aokigahara stealer sample (file name not published)
28099fc5a8cbbd40173f8ac0aded9044fdb4a598  # Aokigahara stealer sample (file name not published)
2cdc77108c585d14b1567053b8f8a58ea78c255f  # Aokigahara stealer sample (file name not published)
2f30b116ca3dc48ff5783c3bb7391d671866e27b  # Aokigahara stealer sample (file name not published)
30737fa46f2ff93e33f45008cf306233954ea898  # Aokigahara stealer sample (file name not published)
324e0fffc7977f8ff591993819123d33c2970953  # Aokigahara stealer sample (file name not published)
3586eeaf3ffe2c3a4f1d005aee863b7aee7cc7bd  # Aokigahara stealer sample (file name not published)
3696baa239c479f9f5103618214972890f1a473b  # Aokigahara stealer sample (file name not published)
385e21a96687a33f79f9cab0627aec7218e1a8f4  # Aokigahara stealer sample (file name not published)
38b2c29b46086b417181deb958e8b276a9d1fbf4  # Aokigahara stealer sample (file name not published)
3aaeb49864108c92af50feeb5d9e3b3efaa39818  # Aokigahara stealer sample (file name not published)
3c56a0c3c59a1425bfdcdffda9e8b308f478b147  # Aokigahara stealer sample (file name not published)
426bea548a24f7f39c279770b5dd21fbed7f7328  # Aokigahara stealer sample (file name not published)
4417d2645adb6e7e94959daf483f93058500840f  # Aokigahara stealer sample (file name not published)
4485feb1385975ef13f08148cdc679f50fa0cc6d  # Aokigahara stealer sample (file name not published)
456414c0acf463e06f95f5bd3d8355758fca6b53  # Aokigahara stealer sample (file name not published)
4595d1875441ffb8c076856e366d1f4f5c74a679  # Aokigahara stealer sample (file name not published)
48c6b66dc4b4d40d72b7460e6fd670b6c03d5d40  # Aokigahara stealer sample (file name not published)
4b3a4ad0acd43fc8796147381074a98e8e576904  # Aokigahara stealer sample (file name not published)
4c07021f8b0cdd3ca334a13f9af2bade9e699dc2  # Aokigahara stealer sample (file name not published)
50060c0fc5fa8932671eb24b03b05d00c2e99951  # Aokigahara stealer sample (file name not published)
51ae6cad54f33846f83a352457b9a44ef8b7a61d  # Aokigahara stealer sample (file name not published)
59bbaca8a848215e7e4af7cec320432c51b0b812  # Aokigahara stealer sample (file name not published)
5ef7eb755f1e3f6bf8586794777466a639057dcb  # Aokigahara stealer sample (file name not published)
5fadf206dc163f63ebf66f984042b00c67121813  # Aokigahara stealer sample (file name not published)
5fe11fbb89d33efa712ffb8590626bd88fecab8f  # Aokigahara stealer sample (file name not published)
61a024f5a42a9a91b0a9871a1613c7fb3ef5613b  # Aokigahara stealer sample (file name not published)
62dcc9ba2a5eee9dd941ad08766603c89a45869b  # Aokigahara stealer sample (file name not published)
6393c0d3fb76f771a9418c2d152c621929ee002b  # Aokigahara stealer sample (file name not published)
63cc5033532cd00bda3d4968b1257629e613505a  # Aokigahara stealer sample (file name not published)
682dd6392004e1131b1aa3f2351a0bc7adb84d14  # Aokigahara stealer sample (file name not published)
699c809a116106da40b3f641c11b0db34d0f1c5e  # Aokigahara stealer sample (file name not published)
6a09fffca381f6624d2def1083284ca486306e7c  # Aokigahara stealer sample (file name not published)
6c860007f8f49de8e9848b538bd7acdc6bfcdf55  # Aokigahara stealer sample (file name not published)
6e073a196a9bfa01da418b5783c2395fb22c7ec9  # Aokigahara stealer sample (file name not published)
6eb8f17dba69f6d25b7f5c529b25d199707b69e2  # Aokigahara stealer sample (file name not published)
6f7d5935602c188b1e3066b4ab9aeef7b85a8ec2  # Aokigahara stealer sample (file name not published)
70915fac07af4ad7acc1868fe05dc00b98206f0f  # Aokigahara stealer sample (file name not published)
70d4007b11e56752fffb81f71c90987f4bb441fa  # Aokigahara stealer sample (file name not published)
70e91644802053cfa68be6ec0f94207282ff27e4  # Aokigahara stealer sample (file name not published)
7233a15327dc2f0cca0292e6352eebf6e6e090d8  # Aokigahara stealer sample (file name not published)
72aa6417a19c45dabdd6c79087fb13c73ae7b823  # Aokigahara stealer sample (file name not published)
7cc6c75fad93b2482a9c5f4522530007c72f99e0  # Aokigahara stealer sample (file name not published)
82ebfc09c98bd211ff8b08efcd8ec8b9ce1fb4be  # Aokigahara stealer sample (file name not published)
83b43ce0ee9881201451c0604fb5b792bd4581f5  # Aokigahara stealer sample (file name not published)
86230c07e638b85bc8010df1aa9e33ca61fda4cc  # Aokigahara stealer sample (file name not published)
89f296aef7988bdd59109e4263d0a31c886770aa  # Aokigahara stealer sample (file name not published)
8b0cf62382f620bbbf5d176295d63ed522ab9c2c  # Aokigahara stealer sample (file name not published)
8c3111a61fbf30e86b8206397ce3c0017e5ea509  # Aokigahara stealer sample (file name not published)
8c4c1d8325adb463c34694ce3a5a851e97b81a6f  # Aokigahara stealer sample (file name not published)
8d7a81119ce7be6eb452192f08fa2669228ff597  # Aokigahara stealer sample (file name not published)
8d860ca64ce0cbf40871fd66b9e3a513e73cda90  # Aokigahara stealer sample (file name not published)
8dc40b7208b8c0d95d1ab8dda9a56fff790b3ddd  # Aokigahara stealer sample (file name not published)
8e9fd1274b77ff9726c4637d008a9d599c9d5f1e  # Aokigahara stealer sample (file name not published)
9003310bf73dc337b8b50131f1c0724a79161aac  # Aokigahara stealer sample (file name not published)
90c057dafefbee423e3014266218272d454a0aa3  # Aokigahara stealer sample (file name not published)
912ed8ce7767f94f62de6a30fb8e9df7c0c512f7  # Aokigahara stealer sample (file name not published)
9369e0e5eae22cff71b375ad285ea89471f25101  # Aokigahara stealer sample (file name not published)
97c80e2d382e6e55f61702da3afb838fdec2f4f1  # Aokigahara stealer sample (file name not published)
98c6649cc256dfadf1af939da46f037b05c0f2e5  # Aokigahara stealer sample (file name not published)
99b28a20851a3ed29769a469f4dd31ebbb8a83de  # Aokigahara stealer sample (file name not published)
9d424ce23753ffddc7b39477222123b0f2c33066  # Aokigahara stealer sample (file name not published)
9f05045f5447d79c8c0d0f5cb589fd85c89620d3  # Aokigahara stealer sample (file name not published)
9f20034441e78574aec7ad39842a342f939eaf6a  # Aokigahara stealer sample (file name not published)
a00da33342a5ea5168d851a5c93109e07cf95c49  # Aokigahara stealer sample (file name not published)
a2abdf1af38da9995a3b6672ce6c34e9e17c39a4  # Aokigahara stealer sample (file name not published)
a4871f4a61faf74e1cceaffe26901f6076c9f37e  # Aokigahara stealer sample (file name not published)
a63ded811d412cd0f859891b0c1f066b87bb1b80  # Aokigahara stealer sample (file name not published)
a84148b4d408544325c20ee04299ae95c762b6a2  # Aokigahara stealer sample (file name not published)
ad960d2abb1999a7260c5f498532a5bb8754c2a5  # Aokigahara stealer sample (file name not published)
ae9d6b1a32a45b05edb920538e538a3d6a5fe439  # Aokigahara stealer sample (file name not published)
af196456d2ebf9556ab239661922565e1bca9ac0  # Aokigahara stealer sample (file name not published)
b335483800d35a29c25767c917725e24c21b0f2b  # Aokigahara stealer sample (file name not published)
b33e59c598845d2547ddc7f654115df2284bb3ae  # Aokigahara stealer sample (file name not published)
b36d0796e33f808eaeb33b33908327a410680c56  # Aokigahara stealer sample (file name not published)
b4491f1919a88deedebe7fba56b593e7d6655458  # Aokigahara stealer sample (file name not published)
b523d56d82485253f2e3619437c8398b7ee0f20d  # Aokigahara stealer sample (file name not published)
b607025d3e06d1e8cffcd398c64871d7fcd92bf4  # Aokigahara stealer sample (file name not published)
b6f0dfd3aa2fc7b8b2ad432c4a4276da23ceb46b  # Aokigahara stealer sample (file name not published)
bab4c910e3c98085bf228519f346d071a75e44e3  # Aokigahara stealer sample (file name not published)
bad339c5ad65808285b0e87498f2ea6f669417d9  # Aokigahara stealer sample (file name not published)
c00aceaa40b38e48f662ac927bb86302934aaf96  # Aokigahara stealer sample (file name not published)
c2e28ba5523c1f3a545a0b4a00a9d5e9c8ae7583  # Aokigahara stealer sample (file name not published)
c323c59815082ac842f0bafcc01f537444e1cc5a  # Aokigahara stealer sample (file name not published)
c5b50b5e14dfb5e028dfd88a4e4fb960dd650b52  # Aokigahara stealer sample (file name not published)
c810acdf30bcdfc6e8021e7f62a3bcce0828c2ee  # Aokigahara stealer sample (file name not published)
c8456bb71a74ed6305455c318295b8a39e9277f3  # Aokigahara stealer sample (file name not published)
cbdff0df0dbdbc3c1845d208661dd75fd335ef93  # Aokigahara stealer sample (file name not published)
cc12fcb9e5e6f7afde66f5c3452ed6e3cd6d5390  # Aokigahara stealer sample (file name not published)
cc708ae3d7aa084279122c0db7250096f750d6b0  # Aokigahara stealer sample (file name not published)
d34491da5aa159314129c0ff67f957d873703e35  # Aokigahara stealer sample (file name not published)
d3c90275dd7279a791d717676be15789376f7aa7  # Aokigahara stealer sample (file name not published)
d8bfff5139b54b6f98f07c051cd25d7aa465fec8  # Aokigahara stealer sample (file name not published)
da986a3cb2ce67456ef374fbdd957d370a1a4367  # Aokigahara stealer sample (file name not published)
daef439e1f110d624ce50fe0f9d9d24f61f7497b  # Aokigahara stealer sample (file name not published)
db3332ce75676389b79cf3603d81cb231cb7c19c  # Aokigahara stealer sample (file name not published)
db69c7c930fd5dc5097847979a965fdea4171d5b  # Aokigahara stealer sample (file name not published)
dba5fef27b53ff479a4d7046cd1f22710081a16c  # Aokigahara stealer sample (file name not published)
de9d152010819c8fb5cb611a4de920399e69cfeb  # Aokigahara stealer sample (file name not published)
dfd1540c0185e74ec65acfcae1af7ba0da36c774  # Aokigahara stealer sample (file name not published)
e052ff49190f54e0ffc52f7b3f7b5e7b9e121c4c  # Aokigahara stealer sample (file name not published)
e3af6d9e3b35e5e04feda3cde96256dd3b099624  # Aokigahara stealer sample (file name not published)
e4d93fa1b2d0fd26e8b877511ea33badb4628169  # Aokigahara stealer sample (file name not published)
e5c43abd1c212b2f2a61d5e1cb7821406ca5dc8e  # Aokigahara stealer sample (file name not published)
e8d09eb2e5857fde39b44aeb48078f5396809026  # Aokigahara stealer sample (file name not published)
ec7cd90d56e5b73caa9d4eb0472bfee68d716121  # Aokigahara stealer sample (file name not published)
eec7ece53d47e50940584f4dec684a24a78f42f2  # Aokigahara stealer sample (file name not published)
f0b23f420659babb3a730b122259440091c53c12  # Aokigahara stealer sample (file name not published)
f30e9730db0663f87c67a4b9a0cf12ee59b1621a  # Aokigahara stealer sample (file name not published)
f35e6d2936c0f6dc26bdc68f6d1599cf438a863a  # Aokigahara stealer sample (file name not published)
f4184e1bb61e19d190555c50fdb9d4591767bf53  # Aokigahara stealer sample (file name not published)
f47d45eea1c42b066f8537baf72ffa266646bf62  # Aokigahara stealer sample (file name not published)
f5993e64a18e7494b25a185de9f63a1df961b99a  # Aokigahara stealer sample (file name not published)
f633fc9fd16dadca2725d3d54aaf13ea1a62174f  # Aokigahara stealer sample (file name not published)
f6d950a5a8e5ba336f748b66745c6c1f189d3362  # Aokigahara stealer sample (file name not published)
f8cf33b18e9932d285696fee40a7fee2a87341c2  # Aokigahara stealer sample (file name not published)
f8dbc552c89bef758e4fcbd9942d06bc0c11db0a  # Aokigahara stealer sample (file name not published)
f990bf4f05dfa4d86b3e3bc0a54f62e91b83f2c6  # Aokigahara stealer sample (file name not published)
fa61cf3bebefd88ea9d06007a4b85791aa851c8a  # Aokigahara stealer sample (file name not published)
fbe8708247fabd8a1bc3963db3911b5e77728669  # Aokigahara stealer sample (file name not published)
fe6b7d6fa8c2424d14a0d465ab78aaf83db0d07b  # Aokigahara stealer sample (file name not published)
ff402e64bc42222c8a4ee703123903c2a2117dd9  # Aokigahara stealer sample (file name not published)

SHA256s

(not available)  # all Aokigahara samples above: source publishes SHA-1 only

Aokigahara clipper component

The SHA-1 comes from the source article; it publishes no MD5 or SHA-256 for this file, and they were not pivoted through OTX or VirusTotal for this report.

MD5s

(not available)  # clipper component -> dropped beneath C:\Microsoft\Vault

SHA1s

8d6c86e0af7807f9b5baf15e9f076bb81e012515  # clipper component -> dropped beneath C:\Microsoft\Vault

SHA256s

(not available)  # clipper component -> dropped beneath C:\Microsoft\Vault

References Appendix

• “Infostealer riding the AI Train”, Check Point Research, Oct. 23, 2026, https://research.checkpoint.com/2026/infostealer-riding-the-ai-train/ (source article; capabilities, targeted paths, anti-analysis, configuration, network protocol, IOCs, YARA rule, SHA-1 hashes)
• MITRE ATT&CK Enterprise technique pages linked above, MITRE, https://attack.mitre.org/