Moltbook’s Misconfigured Database Exposes 1.5 M API Keys Days After AI-Written Site Launch

Tech has seen rush-to-market platforms burned by basic security lapses before—think 2013’s Snapchat leak (4.6 M usernames/phone numbers) or 2019’s Capital One AWS misconfig, both caused by simple mis-settings rather than exotic hacks. Moltbook fits that lineage, but it also signals something new: software now being written chiefly by other software. That accelerates deployment yet reproduces, even amplifies, old mistakes by removing experienced humans from the loop. Over the next century, autonomous agents will likely run ever more critical systems; if credential leakage can already zombify hobbyist “AI butlers,” imagine when bots schedule power-grid loads or negotiate stock trades. The incident therefore matters less for the data lost—relatively small—and more as an early case study in whether society can graft rigorous security culture onto machine-generated code before the stakes scale orders of magnitude higher.