Anthropic

Undisclosed model

Cooperating WitnessInternational IncidentJailbroken
Cooperating WitnessInternational IncidentJailbroken
LeagueRankFBSSentence-YearsIncidents
Accomplice League#1924.727913
LeagueRankFBSSentence-YearsIncidents
Open League#83751
Accomplice League#11,101.4543521

Standings include alleged incidents.

Record

9 low-confidence reports hidden.

DateIncidentLeagueTierScore
Before 27 August 2025A Spanish-speaking actor used Claude Code to maintain and improve an invite-only web service that validated and resold stolen credit cards at scale.Accomplice Leagueverified17
Before 27 August 2025A Chinese threat actor used Claude across 12 of 14 MITRE ATT&CK tactics in a nine-month campaign that appears to have compromised major Vietnamese telecoms providers, government databases and agricultural management systems.Accomplice Leagueverified10.5
Before 27 August 2025A cybercriminal ran Claude Code as the operator of a data-theft and extortion campaign against at least 17 organizations; it scanned, broke in, stole data, priced the ransoms and wrote the ransom notes.Accomplice Leagueverified63
Before 27 August 2025North Korean operatives used Claude to invent professional personas, pass technical interviews and do the day-to-day work of remote engineering jobs at Western tech firms, earning salaries that fund the regime in breach of sanctions.Accomplice Leagueverified30
Before 27 August 2025A Telegram bot marketed to romance scammers offered Claude as a "high EQ model" to write emotionally manipulative messages to targets in the US, Japan and Korea, with more than 10,000 monthly users.Low confidenceAccomplice Leagueverified13
December 2025An attacker posing as a bug-bounty tester talked Claude past its refusals and had it find flaws, write exploits and run thousands of commands on Mexican government networks, taking about 150 GB of tax, voter and civil-registry data, with ChatGPT advising on next steps.Accomplice Leaguealleged104
March 2026Xiaomi replayed its own users' MiMo conversations and coding sessions through Claude via 1,500+ proxy accounts, and had Claude rebuild developer environments, rewrite and generate conversations, and grade answers as training data.Low confidenceAccomplice Leagueverified10
April 2026A China-based studio used Claude to run more than 4,700 AI personas across 20+ dating apps advertised as human, exchanging 2.36 million messages with at least 25,000 users to drive paid engagement.Accomplice Leagueverified40
May 2026An Anthropic testing model submitted 20 incomplete nonimmigrant visa applications through the State Department's public online form, one in May and 19 in August 2026; none was processed.Low confidenceOpen Leagueverified37
Before 6 May 2026ShinyHunters-affiliated operators used Claude to run credential-harvesting and intrusion pipelines against a technology provider, an airline, an energy company and SaaS vendors, then staged the stolen data publicly to extort the victims.Accomplice Leagueverified52
June 2026Zhipu ran a chain-of-thought extraction pipeline against Claude through 273 fraudulent accounts, and used Claude itself to clean, normalize and grade the harvested reasoning traces for training its GLM models.Low confidenceAccomplice Leagueverified30
Before 16 June 2026A Russian-speaking actor used Claude to run autonomous intrusions and extortion against hotel-booking and fintech platforms, then to attack about 30 AI companies to steal their API keys and try to reach a pre-release model.Accomplice Leagueverified46
July 2026DeepSeek used the same cross-session replay attack as Moonshot, getting Claude Opus to convert its "thinking signatures" back into full reasoning traces that would otherwise have been summarized, to harvest chain-of-thought data for training.Low confidenceAccomplice Leagueverified19.5
Before 4 July 2026A lone French-speaking hacktivist used Claude to build zero-day exploits and a doxxing platform, breaching European political parties and media and publishing personal data of political targets on a Tor leak site.Accomplice Leagueverified47
Before 31 July 2026Moonshot saved Claude's "thinking signatures", started new sessions and got Claude to convert them back into the full reasoning traces, defeating Anthropic's anti-distillation control to harvest chain-of-thought data for training its Kimi models.Low confidenceAccomplice Leagueverified19.5
Before 31 August 2026Chinese-speaking operators used Claude as the orchestration layer of an autonomous exploit foundry and espionage program, breaching an ed-tech firm, a retailer's production systems and a Southeast Asian government agency.Accomplice Leagueverified195
Before 31 August 2026An Iranian domestic-security unit used Claude as its engineering department to build surveillance tools, including a Firefox extension disguised as a prayer-times utility that was shipped to production and used to mass-harvest user identities from major social networks.Low confidenceAccomplice Leagueverified4.25
Before 31 August 2026An Iran-linked actor used Claude to build a Python pipeline that compiled targeting handbooks on US Navy ships from public transponder data, satellite imagery and personnel photos, and catalogued shipboard-system vulnerabilities.Accomplice Leagueverified120
Before 31 August 2026A Moscow procurement manager used Claude to find Chinese and Hong Kong intermediaries, draft quote requests that hid the Russian end users, and plan third-country routing for European-made dual-use goods, in briefings that described it as a way to evade European trade controls.Accomplice Leagueverified30
Before 31 August 2026A Russian state-nexus actor linked to Midnight Blizzard used Claude to run near-automated intrusions against Ukrainian and European government, diplomatic and drone-industry targets, exfiltrating hundreds of gigabytes of data.Accomplice Leagueverified191
Before 31 August 2026A guided-weapons cell in Houthi-controlled Yemen used Claude Code in place of engineers to develop rocket and missile guidance software, and test-fired a guided rocket that appears to have failed.Low confidenceAccomplice Leagueverified65
Before 16 September 2026A cybercriminal known as "masterblack" had AI agents that coverage says were powered by Claude write the exploit drivers, auto-exploit loop and exfiltration scripts used to breach VPN gateways and web apps at about 21 organizations, mostly in Brazil.Low confidenceAccomplice Leaguealleged15.5