Skip Navigation
53 comments
  • I simply see it as history repeating itself with prompt hijacking becoming the new code injecting. In that case the fact nothing was changed is poetic. We're back full circle.

  • So are they loading every exam in its entirety into the same context window and doing the whole thing with a single prompt? Or letting the LLM take arbitrary actions instead of only returning a value? It seems like they would have had to make some pretty bad decisions for this to be possible, beyond maybe just adjusting the one grade. I wonder what the prompt jailbreak equivalents of sql injection would actually look like

53 comments