[Anthropic Releases Threat Intelligence Report, Discloses Multiple Cases of AI Misuse] According to an official report from Anthropic, their team disclosed several malicious activities involving the use of Claude between December 2025 and August 2026. These activities include state-sponsored cyber espionage, data theft and extortion groups (such as those associated with ShinyHunters), AI supply chain attacks, and seven other categories of risks. Some attackers attempted to steal corporate AI keys to launch attacks and sought to obtain unreleased Claude models, but all attempts were unsuccessful, and Anthropic's systems remained uncompromised.