News
Twitter pranksters derail GPT-3 bot with newly discovered “prompt injection” hack By telling AI bot to ignore its previous instructions, vulnerabilities emerge.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback