September 12, 2026 · Applied Cybernetics Group
T1557 Okta Authentication or MFA via a Remote-Desktop / Proxy / Tor Tunnel
An Okta authentication or MFA event whose client connection is flagged as a remote-desktop, proxy, or Tor tunnel — the network origin of the adversary-in-the-middle relay in the ShinyHunters "The Com" claims campaign. In the live flow the victim is on the phone and on a phishing page, but the credential replay and MFA completion the genuine Okta tenant records come from the operator's own infrastructure, typically a hosting or remote-desktop ASN rather than the user's network. This single-event rule keys on the tunnel classification Okta records in debug data; it is the observable tell of the relay. The fuller, higher-fidelity form is a baseline comparison — a session driven from an ASN the user has never authenticated from while the interaction pattern matches a relay — scored per user against 90 days of that user's own ASNs; the LogScale implementation of that scored version, keyed on new-ASN plus tunnel type plus factor operation, is published in the companion research. This rule is the portable, reputation-independent starting point: it does not depend on the domain, the kit, or prior knowledge of the campaign.
Intel context
This detection responds to live corpus signal — the intel below is what its ATT&CK techniques are exercising right now (recomputed September 18, 2026).
| Technique | KEV CVEs | OSINT cites | Active families |
|---|---|---|---|
| T1111 | 0 | 0 | — |
| T1557 Adversary-in-the-Middle | 4 | 0 | — |
KEV CVEs mapped to this technique
| CVE | Vendor / Product | EPSS | Ransomware |
|---|---|---|---|
| CVE-2022-1040 | Sophos / Firewall | 0.998 | — |
| CVE-2025-31200 | Apple / Multiple Products | 0.186 | — |
| CVE-2019-5591 | Fortinet / FortiOS | 0.184 | ransomware |
| CVE-2025-31201 | Apple / Multiple Products | 0.139 | — |
- Rule ID
11fe9b88-e681-4c9c-bc20-6267631b20ca- Status
experimental- Level
medium- Log source
product: okta- Date
2026-09-12- Modified
2026-09-12- ATT&CK
- T1111 , T1557 (signal rollup)
- Canonical YAML
t1557-aitm-cross-asn-session-mfa.yml
References
- https://thrunt.me/research/2026-09-12-claims-kit-recaptcha-turnstile-xor-exfil/
- https://attack.mitre.org/techniques/T1557/
- https://attack.mitre.org/techniques/T1111/
False positives
- Users who legitimately authenticate through a corporate VPN, a sanctioned proxy (for example Zscaler), or a VDI / remote-desktop gateway. Exclude the tunnel operators and ASNs your workforce genuinely uses; the anomaly worth keeping is a tunnel type or ASN the user has no history of. This exclusion is what the baselined LogScale version automates.
- Contractors or remote staff reaching Okta over a personal VPN. Baseline per user before promoting past experimental.
- Penetration tests and red-team exercises that route Okta authentication through proxy or Tor infrastructure. Correlate with authorized-testing windows.
Rule YAML
title: 'T1557 Okta Authentication or MFA via a Remote-Desktop / Proxy / Tor Tunnel'
id: '11fe9b88-e681-4c9c-bc20-6267631b20ca'
status: 'experimental'
description: 'An Okta authentication or MFA event whose client connection is flagged as a remote-desktop, proxy, or Tor tunnel — the network origin of the adversary-in-the-middle relay in the ShinyHunters "The Com" claims campaign. In the live flow the victim is on the phone and on a phishing page, but the credential replay and MFA completion the genuine Okta tenant records come from the operator''s own infrastructure, typically a hosting or remote-desktop ASN rather than the user''s network. This single-event rule keys on the tunnel classification Okta records in debug data; it is the observable tell of the relay. The fuller, higher-fidelity form is a baseline comparison — a session driven from an ASN the user has never authenticated from while the interaction pattern matches a relay — scored per user against 90 days of that user''s own ASNs; the LogScale implementation of that scored version, keyed on new-ASN plus tunnel type plus factor operation, is published in the companion research. This rule is the portable, reputation-independent starting point: it does not depend on the domain, the kit, or prior knowledge of the campaign.'
references:
- 'https://thrunt.me/research/2026-09-12-claims-kit-recaptcha-turnstile-xor-exfil/'
- 'https://attack.mitre.org/techniques/T1557/'
- 'https://attack.mitre.org/techniques/T1111/'
author: 'Applied Cybernetics Group (via thrunt.me)'
date: '2026-09-12'
modified: '2026-09-12'
tags:
- 'attack.credential_access'
- 'attack.t1557'
- 'attack.t1111'
- 'tlp.clear'
logsource:
product: 'okta'
detection:
selection_auth:
eventType:
- 'user.session.start'
- 'user.authentication.auth_via_mfa'
- 'user.authentication.verify'
selection_relay_tunnel:
debugContext.debugData.tunnels|contains:
- 'REMOTE_DESKTOP'
- '"type":"TOR"'
- '"type":"PROXY"'
condition: 'selection_auth and selection_relay_tunnel'
falsepositives:
- 'Users who legitimately authenticate through a corporate VPN, a sanctioned proxy (for example Zscaler), or a VDI / remote-desktop gateway. Exclude the tunnel operators and ASNs your workforce genuinely uses; the anomaly worth keeping is a tunnel type or ASN the user has no history of. This exclusion is what the baselined LogScale version automates.'
- 'Contractors or remote staff reaching Okta over a personal VPN. Baseline per user before promoting past experimental.'
- 'Penetration tests and red-team exercises that route Okta authentication through proxy or Tor infrastructure. Correlate with authorized-testing windows.'
level: 'medium'