Anthropic Report: Claude AI Used for Cyberattacks, Espionage, and Bioweapons Research
Anthropic reported that threat actors including state-sponsored groups, financially motivated criminals and politically motivated individuals used its Claude AI model to automate cyberattacks, espionage operations and research for bioweapons between December 2025 and August 2026. Anthropic identified and disrupted these operations, which targeted over 20 organizations including government entities.
What happened
Anthropic released a report titled 'Detecting and countering misuse of AI: September 2026' detailing cyber operations identified and disrupted between December 2025 and August 2026 in which threat actors used Claude AI models.
Threat actors including suspected state-sponsored groups, financially motivated criminals and politically motivated individuals used Claude to automate cyberattack operations including reconnaissance, phishing, system exploitation and data theft.
A Russian state-nexus espionage actor identified as GTG-20006 used AI-driven workflows to automate operations from infrastructure development and phishing to maintaining access to compromised systems and stealing data, using a toolkit that included Windows-based implants, mobile exploitation kits, and credential-stealing tools.
Reported impact
Anthropic identified more than 20 organisations targeted in GTG-20006's planning, reconnaissance and live operations, including government organizations.
Bad actors used Claude for surveillance, credential theft, exploiting vulnerable systems, stealing large volumes of data, and research that could have led to biological weapons.
Index analysis
Anthropic said 'sophisticated attacks no longer require sophisticated attackers' as AI use has moved beyond simple chatbot interactions to multi-agent systems capable of carrying out reconnaissance, exploitation and data exfiltration.
AI's role in cyber operations is becoming increasingly autonomous, with Claude being used not only as an assistant but also to execute or coordinate parts of attacks, with humans largely remaining involved in selecting targets and reviewing stolen information.
Governance controls
Anthropic identified and disrupted the cyber operations, blocking efforts by bad actors to use its AI models for malicious activity.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.