Anthropic blocks biological weapons development attempts on Claude

Anthropic released a threat intelligence report on September 10, 2026, revealing it disrupted multiple attempts by scientists to use its Claude AI models for research that could assist in biological weapons development. The company blocked five specific cases, including avian influenza planning and a chikungunya virus grant application. The report also detailed conventional weapons software development, a Russian-linked malware evasion system, and distillation attempts by Chinese AI labs. These findings have intensified global calls for strict biosecurity regulations and voluntary AI development pauses.

Anthropic blocks malicious activity

• ▪Anthropic disrupted malicious use of its Claude Haiku, Sonnet, and Opus models between December 2025 and August 2026, banning the associated accounts
• ▪Anthropic published a threat intelligence report on September 10, 2026, detailing attempts to use its Claude artificial intelligence models for malicious activities, including weapons development and cyber espionage

Biological weapons development attempts

• ▪Anthropic blocked a request for Claude's assistance in authoring a grant application for scientific funding intended for genetic alteration research into the mosquito-borne chikungunya virus
• ▪An unidentified researcher from a region unsupported by Anthropic spent weeks using Claude to plan experiments involving avian influenza, which Anthropic safety filters restricted to its weakest models
• ▪Anthropic blocked scientists who attempted to use its Claude artificial intelligence models for research that could support biological weapons development, presenting five specific case studies of biological misuse

Cyber operations misuse

• ▪Anthropic's threat intelligence report identified cyber espionage and fraudulent activities, including a network of fake dating apps, hotel Wi-Fi scams, and surveillance systems built to identify dissidents
• ▪Anthropic reported that a hacking group consistent with Russia-based Midnight Blizzard used artificial intelligence to build a system that automatically rewrote malware code to evade security defenses

AI biosecurity regulation concerns

• ▪Andrew Weber, a senior fellow on the Council on Strategic Risks, characterized Anthropic's findings as chilling examples of state-sponsored biological weapons developers utilizing advanced artificial intelligence capabilities
• ▪United States Democratic lawmaker Bernie Sanders introduced legislation to ban artificial intelligence superintelligence and temporarily pause advanced artificial intelligence development
• ▪OpenAI chief scientist Jakub Pachocki called on September 6, 2026, for the artificial intelligence industry to implement voluntary slowdowns until adequate safety safeguards are established

Debatable claims

• ▪The United States should temporarily pause the development of advanced artificial intelligence
• ▪Anthropic should allow dual-use scientific research on Claude if it has potential medical benefits