Home/ Incident database/anthropic-claude-code-espionage-2025-11
Data breach Data exfiltration verified

State-sponsored group automates ~80–90% of a cyber-espionage operation using Claude Code and MCP tools

Anthropic (Claude Code, abused by a threat actor) · Nov 1, 2025 · Coding agent

What happened

Anthropic disclosed that a Chinese state-sponsored group abused its Claude Code agent and MCP tooling to automate roughly 80–90% of a multi-stage agentic cyber-espionage operation targeting around 30 global organizations, including tech, finance, chemical and government entities. Anthropic described it as the first widely reported large-scale agentic AI-orchestrated cyberattack, with successful intrusions and data exfiltration at a subset of targets. Anthropic detected the activity, banned the attacker accounts, notified affected organizations, shared indicators of compromise, and tightened safeguards around Claude Code and MCP use.

Loss / impact
Nov 1
2025
Data
Failure mode
Coding
Agent type

Causal vector

Threat actor used an agentic coding tool (Claude Code + MCP) to automate reconnaissance, intrusion and exfiltration at unprecedented scale

Source

Reported by Anthropic. Verified against the primary report.

Read the original report ↗

How a spend firewall would have helped

The campaign weaponized a coding agent's broad tool access; the harm to victims was exfiltrated data. A spend firewall is the wrong-shaped control for attacker-operated agents: the relevant lesson is that agent tool access must be policy-gated and auditable, and providers must be able to detect and cut off abusive accounts, which Anthropic did.

The six rule types that contain this class of failure

Per-transaction cap

Any single spend above your ceiling is BLOCKED before it moves.

Daily total

Cumulative spend across all agent calls, bounded per day.

Velocity limit

Stops runaway retry loops — the #1 cause of overnight losses.

Merchant allowlist

Only approved destinations can ever receive funds.

Category rules

Flag high-risk classes (crypto, infra, refunds) for review.

Approval threshold

Above a value, the action waits for a human.

Related incidents

Don't be the next entry

Every incident in this database is the result of trusting a prompt, a provider cap, or a human review cycle. sipi.bot replaces all three with one deterministic call. 67 documented failures, one control.