Home / Mar 16, 2026 / Story
0
#8 The Hacker News general March 14, 2026 at 16:17 UTC

OpenClaw AI Agent Flaws Could Enable Prompt Injection and Data Exfiltration

By [email protected] (The Hacker News)

AI Summary

China's CNCERT warned about security flaws in OpenClaw (formerly Clawdbot/Moltbot), an open-source AI agent platform with weak default configurations that enable prompt injection and data exfiltration attacks. The vulnerabilities affect organizations deploying autonomous AI agents and highlight the security risks of self-hosted AI systems with insufficient access controls.

Relevance score: 68.0/100

# More from March 16