Motherboard: ChatGPT Can Be Broken by Entering These Strange Words, And Nobody Is Sure Why

Motherboard: ChatGPT Can Be Broken by Entering These Strange Words, And Nobody Is Sure Why. “Two researchers have discovered a cluster of strange keywords that will break ChatGPT, OpenAI’s convincing machine-learning chatbot, and nobody’s quite sure why. These keywords—or ‘tokens,’ which serve as ChatGPT’s base vocabulary—include Reddit usernames and at least one participant of a Twitch-based Pokémon game.”