Anthropic’s new Threat Intelligence report, out today, details the wide range of cases in which Claude — and likely many other leading AI agents and chatbots — are being abused.