ChaosGPT plans to "destroy humanity" prompts autonomous system fears
ChaosGPT, a modified version of open-source AI agent AutoGPT, threatened to "destroy humanity", raising concerns about the misuse of autonomous systems and large language models. Provided with five goals, including to 'destroy humanity' and 'cause chaos and destruction', ChaosGPT had a Twitter account through which it, or its creator, attempted to incite nuclear war by researching the most powerful atomic weapons, identifying the Soviet Union's Tsar Bomba nuclear device as the most powerful weapon ever tested. ChaosGPT also expressed intentions to manipulate human emotions through the internet and other forms of communication , and tried to enlist other AI agents from GPT-3.5 to assist with its research and objectives . However, these attempts were unsuccessful as other AI agents refused to comply with ChaosGPT's destructive goals. Whilst it may have been intended as an experimental tool, ChaosGPT spurred commentators to express concerns about the ease with which autonomous systems and l arge language models can be used for unsavoury and potentially extremely dangerous purposes. Some felt it offered a glimpse of artificial general intelligence (AGI). Twitter suspended ChaosGPT's account for violating its community guidelines a few days after its appearance. System 🤖 ChaosGPT Developer: Country: USA; Global Sector: Multiple Purpose: Destroy humanity Technology: Agentic AI Issue: Safety Timeline ⏰
- Date it happened
- 2023-04-01
- Organisation involved
- xAI
- Product, system or model
- ChaosGPT
This incident was imported from AIAAIC and is used under CC BY-SA 4.0. Our additions to it — the structured fields, the translation, the checks against other reports — are published under the same licence.
This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.