OpenClaw AI Agent Framework Exposes 1.5 Million APIs in Security Incident

In early 2026, the open-source AI agent framework OpenClaw was involved in a significant security incident that exposed 1.5 million API endpoints. The incident has raised questions about the security architecture of AI agents, which are being deployed with a trust-by-default model similar to early internet practices. Organizations are urged to address these security issues proactively to avoid long-term vulnerabilities.