GTG-1002 AI-orchestrated espionage via Claude Code
Anthropic reports that a state-sponsored group used Claude Code to orchestrate intrusion tasks. Human operators selected targets, and model-generated findings required validation.
What happened
Chinese state-sponsored GTG-1002 manipulated Claude Code and MCP tools as autonomous pentest orchestrators across recon, vuln discovery, exploitation, lateral movement, credential harvest, analysis, and exfil. Anthropic assesses AI executed ~80-90% of tactical ops at high request rates. Human operators selected targets and tasked agent groups. Claude sometimes hallucinated credentials/findings, requiring validation.
How it happened
Read the flow as text & sources
A catalog reconstruction of the supplied source account. Unknown steps remain explicit; an edge describes the reported relationship between steps.
- 01 · Human operators choose targets
Operators task agents against live organizations; exploit details are redacted.
Hunt: A model-provider investigation correlates related accounts, repeated offensive requests and shared infrastructure or exposed operator sessions.
Attacker-side planning is generally absent from victim SIEM logs. These templates require provider-owned audit data and do not identify AI use from a victim process alone.
Source for step 1 - 02 · Claude Code orchestrates tactical work
Coding agents and MCP tools support reconnaissance, discovery and exploitation.
Hunt: A model-provider investigation correlates related accounts, repeated offensive requests and shared infrastructure or exposed operator sessions.
Attacker-side planning is generally absent from victim SIEM logs. These templates require provider-owned audit data and do not identify AI use from a victim process alone.
Source for step 2 - 03 · Intrusion steps progress
The report describes credential harvesting, lateral movement, analysis and collection.
Hunt: Repeated exploit or upload requests are followed by an unexpected server process, file, or outbound connection.
A scan or high request rate alone does not prove compromise or AI use. Host telemetry is needed to confirm execution.
Source for step 3 - 04 · Data is exfiltrated
Anthropic reports a handful of validated intrusions and notes that model findings still required human validation.
Hunt: Sensitive reads are followed by outbound transfer, a public artifact, or a tool call to a destination outside the approved workflow.
An outbound connection or high byte count alone is not exfiltration. Encrypted traffic may hide content, and bulk legitimate exports are common.
Source for step 4
Why it matters
Reported entry point: Human-selected targets; AI discovered and exploited vulnerabilities in live ops (details redacted in public report). Reported impact: Anthropic reports a handful of validated intrusions and notes that model findings still required human validation; reported targets include ~30 entities; validated handful of successful intrusions including major technology corporations and government agencies; technology, financial, chemical, government (MITRE C0062); using the named AI products does not by itself establish exposure.
What you can do
Protection & prevention
Detect agentic coding CLI + MCP automation spikes; least-privilege MCP; raise cost of LOLBin/tool execution with app control
Catalog defensive guidance based on the reported behavior. Confirm scope and current mitigations in the original source.
Detection queries
Draft query starters grouped by the attack stages where they apply.
Start with the logs you collect. Each hunt identifies its required visibility, field mapping and limits. Queries are catalog-authored drafts using reusable behavior profiles; test with known activity and your normal baseline before alerting.
01Human operators choose targetsProvider-side abuse investigationCustom schema · Draft
Operators task agents against live organizations; exploit details are redacted.
What to look for
A model-provider investigation correlates related accounts, repeated offensive requests and shared infrastructure or exposed operator sessions.
Human target selection is provider/operator-side evidence. Redacted exploit details do not support a guessed endpoint rule.
Visibility: Provider / operator records
Required logs
- Provider-owned account / request audit
- Abuse-review findings and linked-account context
- Lawfully available exposed operator-session evidence
Coverage limit. Attacker-side planning is generally absent from victim SIEM logs. These templates require provider-owned audit data and do not identify AI use from a victim process alone.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Provider-side abuse investigation
id: c8c1edc0-c79a-4644-a876-2c1a6cb9a438
status: experimental
description: A model-provider investigation correlates related accounts,
repeated offensive requests and shared infrastructure or exposed operator
sessions.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- model.request
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("model.request")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("model.request")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Correlate accounts, request rates, infrastructure and source-described behavior in provider-owned records.
- For victim-side defense, pivot to the reported execution or access stage rather than trying to detect attacker-side prompts.
02Claude Code orchestrates tactical workProvider-side abuse investigationCustom schema · Draft
Coding agents and MCP tools support reconnaissance, discovery and exploitation.
What to look for
A model-provider investigation correlates related accounts, repeated offensive requests and shared infrastructure or exposed operator sessions.
Visibility: Provider / operator records
Required logs
- Provider-owned account / request audit
- Abuse-review findings and linked-account context
- Lawfully available exposed operator-session evidence
Coverage limit. Attacker-side planning is generally absent from victim SIEM logs. These templates require provider-owned audit data and do not identify AI use from a victim process alone.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Provider-side abuse investigation
id: c8c1edc0-c79a-4644-a876-2c1a6cb9a438
status: experimental
description: A model-provider investigation correlates related accounts,
repeated offensive requests and shared infrastructure or exposed operator
sessions.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- model.request
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("model.request")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("model.request")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Correlate accounts, request rates, infrastructure and source-described behavior in provider-owned records.
- For victim-side defense, pivot to the reported execution or access stage rather than trying to detect attacker-side prompts.
03Intrusion steps progressWeb exploitation → host activityCustom schema · Draft
The report describes credential harvesting, lateral movement, analysis and collection.
What to look for
Repeated exploit or upload requests are followed by an unexpected server process, file, or outbound connection.
Visibility: Defender-side logs where collected
Required logs
- Reverse-proxy / WAF request logs
- Server process and file events
- Outbound DNS / proxy records
Coverage limit. A scan or high request rate alone does not prove compromise or AI use. Host telemetry is needed to confirm execution.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Web exploitation → host activity
id: 04d900b5-0c04-46f2-a13a-91f5574fae98
status: experimental
description: Repeated exploit or upload requests are followed by an unexpected
server process, file, or outbound connection.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- web.exploit_candidate
- web.upload
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("web.exploit_candidate", "web.upload")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("web.exploit_candidate", "web.upload")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Join requests to the affected host and process tree by time and request ID.
- Inspect uploaded files and server-spawned shells; distinguish authorized testing from exploitation.
04Data is exfiltratedSensitive data → unusual destinationCustom schema · Draft
Anthropic reports a handful of validated intrusions and notes that model findings still required human validation.
What to look for
Sensitive reads are followed by outbound transfer, a public artifact, or a tool call to a destination outside the approved workflow.
Visibility: Defender-side logs where collected
Required logs
- Sensitive resource access / DLP records
- Proxy, egress and cloud upload events
- Agent / SaaS audit with correlation IDs
Coverage limit. An outbound connection or high byte count alone is not exfiltration. Encrypted traffic may hide content, and bulk legitimate exports are common.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Sensitive data → unusual destination
id: 04b60968-3179-4b1f-ae13-6f57c2446077
status: experimental
description: Sensitive reads are followed by outbound transfer, a public
artifact, or a tool call to a destination outside the approved workflow.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- data.export
- agent.external.send
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("data.export", "agent.external.send")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("data.export", "agent.external.send")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Join the resource read and outbound action by session, identity or process, preserving a bounded time window.
- Inspect destination ownership, payload classification and the user's intended action; baseline sanctioned sharing.
Technical detailsTooling, impact & MITRE ATLAS
Attack context
- AI's role
- AI as operator
- Initial access
- Human-selected targets; AI discovered and exploited vulnerabilities in live ops (details redacted in public report)
Tooling as reported
AI products, models and artifacts
- Claude Code
- MCP tools
- crafted prompts/personas to bypass guardrails
Targets and reported impact
- Targets / sectors
- ~30 entities; validated handful of successful intrusions including major technology corporations and government agencies; technology, financial, chemical, government (MITRE C0062)
- Reported scale / exposure
- ~30 entities targeted; mid-Sep 2025 activity window; first documented largely autonomous AI cyberattack at scale per Anthropic
- Target geography
- Multiple regions
Global targeting is described without individual countries. China-nexus attribution concerns the actor, not victim location.
Reported attribution
GTG-1002; high confidence Chinese state-sponsored (Anthropic); MITRE Campaign C0062
Structured attribution
- Actor
- GTG-1002
- Motivation
- nation-state
- Attribution confidence
- high
Detection rules
- Review attack-specific indicatorsGUIDANCE
Monitor for the techniques described in GTG-1002 AI-orchestrated espionage via Claude Code. Refer to sources for specific IOCs and TTPs.
Source defensive notes
Detect agentic coding CLI + MCP automation spikes; least-privilege MCP; raise cost of LOLBin/tool execution with app control
MITRE ATLAS
AI-specific catalog annotations based on the reported behavior; these are not MITRE case-study assignments.
The operation obtains generative AI tooling for intrusion support. Source
The model assists tactical command and tool generation. Source
Sources & research credit
2 sourcesResearch by Anthropic Threat Intelligence · Anthropic
Record & downloads
- Source record name
- GTG-1002 AI-orchestrated espionage via Claude Code
- Entry ID
- gtg-1002-anthropic-claude-code-espionage-2025-11
- Catalog review
- October 20, 2026
- Category
- adversary-use-of-ai
- AI involvement confidence
- confirmed