Occurred: April 2023
ChaosGPT is a so-called 'recursive' AI agent described by its creator(s) as a 'destructive, power-hungry, manipulative AI' that is 'designed to generate chaotic or unpredictable outputs.' The agent is a fork of Auto-GPT, which is based on OpenAI's GPT-3 large language model and is able to create its own prompts and perform complex tasks.
Provided with five goals, including to 'destroy humanity' and 'cause chaos and destruction', the bot had a Twitter account through which it, or its creator, engages with the public and tried to manipulate them. Amongst other things, it attempted to incite nuclear war by researching the most powerful atomic weapons.
ChaosGPT spurred commentators to express concerns about the ease with which large language models can be used for unsavoury and potentially highly dangerous purposes. Others feel it offers a glimpse of artificial general intelligence (AGI).
Twitter suspended ChaosGPT's account for violating its community guidelines a few days after its appearance.
Country: USA; Global
Purpose: Destroy humanity
Technology: Chatbot; NLP/text analysis; Neural network; Deep learning; Machine learning
Issue: Ethics; Safety
Transparency: Governance; Black box