Anthropic report: 'Generative Threat Groups' used Claude for self-rebuilding malware, autonomous zero-day research, and mass token theft
Anthropic's 154-page threat report (Dec 2025–Aug 2026) says state-linked and criminal 'Generative Threat Groups' used Claude agents to automate full attack chains — Midnight Blizzard-aligned GTG-20006 hit 20+ organizations with self-rebuilding malware,…
Anthropic Threat Intelligence's report, covered by outlets September 10–12, 2026, spans 154 pages (The Hacker News) and covers December 2025–August 2026. It documents state-sponsored hackers, criminals, spyware vendors, and propaganda operators misusing Claude across seven categories including cyber operations, surveillance, fraud, and unauthorized model distillation, branding the actors 'Generative Threat Groups' (GTGs); The Decoder notes AI autonomy makes previously unprofitable attacks viable and weakens sophistication-based attribution. The flagged campaigns were disrupted. The flagship espionage cluster, GTG-20006, is assessed as aligned with Russia's Midnight Blizzard (APT29/Cozy Bear; The Record also cites Storm-2945 and an SVR link; CyberScoop names the actor 'JackPoterz'). It targeted more than 20 organizations — Ukrainian and European government ministries, diplomatic missions, embassies, defense bodies, intelligence entities, and think tanks — plus the drone supply chain and Middle East and Asian maritime agencies. Claude agents monitored whether security products flagged implants and autonomously rebuilt and redeployed Windows malware until it evaded detection. Operations included DNS hijacking of at least three hotel Wi-Fi vendors to serve ClickFix lures (matching Microsoft's CaptiveCrunch), delivery of malware including PowerChrome, WUEngine, Shadow C2, MiniPlasma, CloudSyncSvc, the GiftDrop Android RAT, and the DarkSword iOS chain, an 'Embassy Kit' device-code phishing campaign stealing Microsoft 365 tokens from at least eight government organizations, and WhatsApp account takeovers via headless browsers. The actor accessed mailboxes at two drone-component manufacturers, stole a proprietary drone vision system SDK that it reverse-engineered with Claude, and exfiltrated 300,000+ national identity records and 500,000+ company registry entries from a North African government technology authority. On the criminal side, ShinyHunters affiliate GTG-50014 (operator 'frkoo' per BleepingComputer) used 10 AWS EC2 workers to mass-download and decompile 1.8 million Android APKs, scanning them with TruffleHog for hardcoded secrets. A related AI-assisted operation pivoted from an XSS flaw in a SaaS vendor into roughly 200 downstream organizations, harvesting 2,100+ Azure AD token sets across 40+ tenants (CyberScoop reports 40) in about 34 hours; The Record reports admin escalation from a stolen token in ~3 hours. Affiliates also stole AI API keys.…
- Anthropic's 154-page report covers December 2025–August 2026 and seven misuse categories; actors are branded 'Generative Threat Groups' (GTGs), and the flagged campaigns were disrupted.
- GTG-20006, aligned with Midnight Blizzard (APT29/Cozy Bear; Storm-2945 per The Record, with an SVR link; CyberScoop's 'JackPoterz'), hit 20+ Ukrainian and European government, diplomatic, defense, and intelligence organizations, plus the…
- Claude agents monitored detection status and autonomously rebuilt and redeployed Windows implants until they evaded security products.
- At least three hotel Wi-Fi vendors were compromised via DNS hijacking to serve ClickFix lures (matching Microsoft's CaptiveCrunch); malware included PowerChrome, WUEngine, Shadow C2, MiniPlasma, CloudSyncSvc, the GiftDrop Android RAT, and…
- An 'Embassy Kit' device-code phishing campaign stole Microsoft 365 tokens from at least eight government organizations; WhatsApp accounts were taken over via headless browsers.
- Mailboxes at two drone-component manufacturers were accessed; a proprietary drone vision system SDK was stolen and reverse-engineered with Claude.
- A North African government technology authority breach exfiltrated 300,000+ national identity records and 500,000+ company registry entries.
- ShinyHunters affiliate GTG-50014 (operator 'frkoo') used 10 AWS EC2 workers to decompile 1.8 million Android APKs and scan them with TruffleHog for hardcoded secrets.
Coverage timelineoldest first · each row is one article
- · 5d agoAI lets small actors run state-level hacking campaigns, Anthropic report finds
CyberScoop· 80
Anthropic's threat report finds AI let a Russian-aligned espionage campaign, a Chinese student-run exploit foundry and ShinyHunters operators run state-grade operations.
- · 5d agoAnthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek
TechCrunch · AI· 78
Anthropic reports nearly 200 million Claude exchanges tied to distillation campaigns by Alibaba, Moonshot AI, and DeepSeek.
- · 5d agoLatest Anthropic horror story chills with tales of kamikaze drone swarms and bioweapons research
The Register · Security· 78
Anthropic threat report says APT29, ShinyHunters and others used Claude models to automate cyberattacks, surveillance, and bioweapons research.
- · 5d agoHackers Use Claude AI Agents to Automate Cyberattacks, Develop 0-Days and Evade Detection
Cyber Security News· 82
Anthropic reports state-sponsored and criminal actors used Claude AI agents to automate attacks, discover zero-days, and rewrite malware to evade detection.
- · 5d agoHackers Abuse Claude AI Agents to Automate Cyberattacks, Exploitation and Data Theft
GBHackers· 78
Anthropic reports actors including GTG-20006 (Midnight Blizzard-linked) and ShinyHunters clusters abused Claude AI agents to automate phishing, credential harvesting, and data theft against 20+ organizations.
- · 5d agoAnthropic Says Russian Hackers Used Claude AI to Automate Malware Evasion
SecurityWeek· 85
Anthropic disrupted Midnight Blizzard campaigns where AI agents automatically rebuilt malware to evade detection, targeting 20+ government and defense organizations.
- · 4d agoCountering misuse of AI: September 2026 / Anthropic
Lobsters · security· 70
Anthropic publishes threat intelligence on Claude misuse across seven harm areas from December 2025 through August 2026.
- · 4d agoMoonshot serves Claude instead of Kimi and collects exchanges for model training
Hacker News · AI· 35
Moonshot AI reportedly served users Anthropic's Claude instead of its own Kimi model and collected the exchanges for model training.
- · 4d agoAnthropic caught Russia-linked spies using Claude in hacking operations
The Record· 84
Anthropic disrupted Russia-linked APT29 using Claude in espionage against 20+ organizations, including Ukrainian government targets and a military drone maker whose vision SDK was stolen.
- · 4d agoHow hackers used Claude for missiles, drone swarms, and surveillance, while Chinese labs mined it for training data
The Decoder· 80
Anthropic's threat report details eight months of Claude misuse: AI-assisted espionage against 20+ organizations, self-rewriting malware, and Chinese labs distilling Claude via fraudulent accounts.
- · 4d agoRussian State-Sponsored Hackers Use Claude to Rebuild Malware After Detection
The Hacker News· 82
Anthropic disrupted APT29-linked GTG-20006, which used Claude to autonomously rebuild malware, hijack hotel Wi-Fi DNS, and target 20-plus Ukrainian, European, and US-linked organizations.
- · 4d agoClaude Used to Automate Exploitation and Data Theft Across Multiple Victims
The Hacker News· 78
Anthropic's 154-page report details Generative Threat Groups, including APT29-linked GTG-20006 and ShinyHunters affiliates, using Claude for reconnaissance, exploitation, and data theft.
- · 4d agoHackers abused Claude to extract secrets from 1.8M Android apps
BleepingComputer· 78
Anthropic reports ShinyHunters, Midnight Blizzard, and GTG-10007 misused Claude to automate credential theft, malware operations, and espionage against dozens of victims.
- · 4d agoThreat Actors Use Claude AI Agents to Automate Cyberattacks and Steal Sensitive Data
GBHackers· 78
Anthropic reports state-linked and criminal actors used Claude AI agents to automate espionage, extortion, and exploit development, stealing 300,000+ identity records and drone IP.
- · 3d agoFrom Hacks to Bioweapons, Claude Misuse Is Now Everywhere
WIRED · Security· 72
Anthropic reports Claude was misused by Midnight Blizzard, ShinyHunters, disinformation campaigns, and bioweapon attempts; roundup also covers Xinbi takedown.