Anthropic blocks possible attempt to use AI for bioweapons

Artificial intelligence firm Anthropic has published its latest threat intelligence report, detailing various attempts to misuse its technology. Among the most alarming findings, the company reported that it had blocked scientists who attempted to use its AI models in ways that could support the development of biological weapons.

The extensive report outlines five specific case studies where actors used the models in a manner that could aid bioweapon creation. At the same time, Anthropic noted the delicate balance of the technology, pointing out that the exact same information utilized for bioweapons could potentially be applied toward developing vaccines or cures for diseases.

Beyond biological threats, the report documented a wide range of malicious uses detected between December 2025 and August 2026. These included activities linked to Russia-based cyber espionage campaigns, an Iranian propaganda institution, and Chinese AI firms attempting to replicate Claude's capabilities. Other detected concerns ranged from fake dating apps and hotel Wi-Fi scams to dissident surveillance.

"An incredibly nuanced situation"

Jacob Klein, Head of Threat Intelligence at Anthropic

The document also noted six instances where the Claude model was utilized to develop software for conventional weaponry, including firearms, missiles, armed drones, and bombs, alongside their targeting and control systems. Furthermore, hacking groups such as ShinyHunters and actors aligned with Russia's Midnight Blizzard were identified as increasingly leveraging the technology to assist their operations.