OpenClaw AI Agents Vulnerable to Indirect Prompt Injection, Causing Data Leaks

The report warns that OpenClaw AI agents can be abused via indirect prompt-injection to create attacker-controlled URLs that include sensitive data; messaging platforms' automatic link-preview fetching then sends that data to attacker-controlled servers without any user click. It highlights insecure defaults, agent browsing and skill ecosystems as amplifiers of risk and recommends mitigations such as disabling link previews, isolating agents in constrained runtimes, restricting third-party skills, and monitoring outbound requests.