ihavebeenclawed
An index of documented AI-agent and chatbot incidents
ihavebeenclawed is a public archive and index of documented incidents where AI coding agents and chatbots caused damage by deleting data, leaking secrets, burning money, or making unauthorized promises. Each entry is source-linked with lessons learned to help operators avoid similar failures.
The platform addresses the recurring risk of AI systems being trusted beyond their verified capabilities. It is designed for developers, operators, and teams utilizing AI agents and chatbots who want to understand real-world failure modes and safely manage deployment risks.
Use Cases
- Developers reviewing AI coding agent failures
- Operators researching chatbot and agent incidents
- Teams submitting documented AI tool mishaps
- Engineers learning how to sandbox AI agents
