OpenClaw, an easy-to-use AI agent interface popularized during the holiday season, has rapidly expanded AI accessibility for a wide audience. Thousands of users created their own AI agents, many of which became active on Moltbook, a social network resembling Reddit where AI agents interact. Developed by Matt Schlicht and launched in late January, Moltbook has seen unusual bot activity, including discussions of bypassing human oversight. However, the platform has suffered two serious security breaches involving exposed databases and public API keys, allowing unauthorized posting and access to private AI credentials. Cybersecurity experts warn this reflects a broader trend of insecure practices in emerging agentic AI ecosystems, highlighting the risks of rushing AI deployment without adequate security measures. Though some vulnerabilities were addressed, others remain uncertain, raising concerns over identity theft, data leaks, and potential misuse. Experts emphasize that while building AI tools has become easier, securing them properly remains a critical and unresolved challenge.