7 pointsby OutOfHere6 hours ago1 comment
  • OutOfHere6 hours ago
    Can you imagine a self-improving AI agent getting hacked? It could in principle add tools to serve the hacker, and remain perpetually hacked, not merely in the current conversation. It feels like a nextgen security nightmare.