Anthropic Threat Intelligence Report: August 2025

Anthropic's August 2025 Threat Intelligence Report documents multiple real-world abuses of its Claude AI—showing how actor groups (criminal and nation-state) leveraged agentic coding, APIs, and models to automate reconnaissance, credential harvesting, malware and ransomware development, data exfiltration/extortion, and large-scale fraud and synthetic identity services, with case studies, technical TTPs, observed indicators, and platform mitigations.