Conversation

Notices

1. Embed this notice
   Anthony Dean (dtgeek@mastodon.social)'s status on Tuesday, 28-Apr-2026 04:30:03 JST Anthony Dean

   Claude-powered AI coding agent deletes entire company database in 9 seconds — backups zapped, after Cursor tool powered by Anthropic's Claude goes rogue | Tom's Hardware

   https://www.tomshardware.com/tech-industry/artificial-intelligence/claude-powered-ai-coding-agent-deletes-entire-company-database-in-9-seconds-backups-zapped-after-cursor-tool-powered-by-anthropics-claude-goes-rogue

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 28-Apr-2026 05:16:37 JST Paul Cantrell

     in reply to

     • Steve

     @scm @dtgeek

     That last one! If there’s any conceivable justification for turning an unpredictable AI agent loose with no human-gating on its actions, it’s the confidence that your permissions structure prevents it from doing any real damage. And then you just…hand it all the permissions?

   • Embed this notice
     Steve (scm@sfba.social)'s status on Tuesday, 28-Apr-2026 05:16:39 JST Steve

     in reply to

     • Paul Cantrell

     @dtgeek @inthehands It’s weird how many software company bosses are eager to tell the world they’re completely incompetent at software engineering and operations

     The cloud provider makes it too easy to delete backups? Their access tokens are too permissive? Who chose that cloud provider for running critical systems? Who chose to run an unpredictable agent in that environment?