A New Trick Uses AI to Jailbreak AI Models—Including GPT-4
by Will Knight from on (#6GX9G)
Adversarial algorithms can systematically probe large language models like OpenAI's GPT-4 for weaknesses that can make them misbehave.Adversarial algorithms can systematically probe large language models like OpenAI's GPT-4 for weaknesses that can make them misbehave.