Anthropic says it disrupted cases in which threat actors attempted to use Claude for cyber operations, weapons-related research, surveillance, propaganda and biological work, highlighting the growing security challenges surrounding increasingly capable Artificial Intelligence systems.