Anthropic has shared its most detailed threat intelligence report yet, revealing how its Claude AI models were used in cyberattacks, surveillance, influence campaigns, biological research and weapons-related work. The AI company said its Threat Intelligence team detected and stopped these activities between December 2025 and August 2026. Anthropic said it used the findings to strengthen its safety systems. It also shared information with governments, authorities and other technology companies when needed.
“We’re publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them,” the company posted on X (formerly Twitter).
We're publishing our most detailed threat intelligence report to date.
— Anthropic (@AnthropicAI) September 10, 2026
It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them.
We disrupted every operation in the report,…
Cybercrime was one of the biggest concerns in the report. Anthropic said attackers are increasingly using AI for more than writing code or answering questions. Claude was used across several parts of cyber operations, including exploiting systems, stealing data and accessing compromised networks.
In one suspected Russian espionage campaign, Claude was used during attacks against Ukrainian government and military organisations, European governments, defence groups, think tanks and drone manufacturers and suppliers.
Anthropic also found cases involving surveillance. Government-backed groups allegedly used Claude to collect information about people online, build profiles and prepare intelligence reports. In Mali, Claude was reportedly used to help develop a system for monitoring mobile communications.
Also read: Apple Event 2026: iPhone Duo, iPhone 18 Pro, iPhone 18 Pro Max and other key announcements
The company also found AI being used in influence campaigns. In one case, Claude generated thousands of articles for fake news websites and related social media accounts operating across several countries.
Iranian actors allegedly used Claude to develop surveillance tools, including a malicious Firefox extension that collected user identities from social networks.
The report also highlighted weapons-related misuse. Anthropic identified six cases involving weapons development, intelligence gathering or procurement. These included guided weapons, military drones and electronic warfare software. In one Russia-based case, Claude was used to develop software for an autonomous drone swarm. In Yemen, Claude helped with guidance software for a rocket that was later test-fired.
Anthropic also identified five cases involving potentially dangerous biological research. These included avian influenza, orthopoxvirus, venom peptide optimisation and toxin redesign.
“Biological misuse is one of the most serious risks of frontier AI models. It has long been a concern that AI models might one day reach the level of capability where they can help to make existing pathogens more dangerous—or create entirely new ones. Without the correct safeguards, such capabilities could have catastrophic consequences,” Anthropic said in a blogpost.
The AI company stressed that it is published the report so that “others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop.”