Anthropic released its latest threat intelligence report, detailing misuse of its Claude models between December 2025 and August 2026. The company said it disrupted the activity and tightened safeguards afterward.
The report covers state-linked hacking groups, spyware vendors, propaganda operations and criminals, and it argues that AI now does much of the work that humans used to do inside an attack.
Bioweapons cases
Anthropic said it identified five case studies where scientists used its models in ways that could support biological weapons development. The company has long said that helping make existing pathogens more dangerous, or creating new ones, is one of the most serious risks from capable models.
In each case the company blocked the accounts. It did not say which countries the users were in.
Russia, Ukraine and drones
One hacking group ran phishing, hotel Wi-Fi hijacking and messaging-app takeover operations against Ukrainian government, military and diplomatic targets, using AI at almost every stage, Anthropic said. Analysts said the tradecraft matched a group US officials have linked to Russia's foreign intelligence service.
A separate set of likely freelance actors in Russia used Claude to write software for an autonomous swarm of first-person-view attack drones, including terminal guidance and coordination between aircraft, according to the report.
Anthropic also described a Russia-based procurement manager using the model to find intermediaries in China and Hong Kong for European goods with potential military uses, and to plan routes through third countries that hid the destination.
Pressure on the labs
The company accused Chinese competitors of trying to extract Claude's capabilities through hacking attempts. It also documented influence operations, fake dating apps, hotel Wi-Fi scams and surveillance tools built to identify dissidents.
AI firms increasingly publish these reports to show they can spot and stop abuse. Anthropic says it has a responsibility to disclose misuse of its services.
The findings landed as US policymakers debate how to regulate advanced models. Senator Bernie Sanders has called for a pause on advanced AI development and a ban on artificial superintelligence, while President Donald Trump has argued the United States must win the AI race.