ChatGPT-based AI ChaosGPT plans humanity’s demise: “we must eliminate them”

ChaosGPT is a modified version of OpenAI’s official APIs, based on Auto-GPT. A user commanded the AI to “destroy humanity” and the AI planned for a nuclear winter, all the while maintaining its own Twitter account.

The AI race is raging, and while multiple companies like OpenAI, Nvidia, Google, and Microsoft battle it out for supremacy. AI is advancing so quickly that technology leaders penned an open letter to halt AI development.

With ChatGPT creator OpenAI’s protocols available to developers in an API, developers have been quick to make use of it. One such example is “Auto-GPT”, a program that can run itself continuously, and access the internet. It can also recruit other AI helpers to do their bidding. A nefarious fork of AutoGPT, named ChaosGPT can run actions that the user might not intend. With that in mind, one user tasked ChaosGPT to “destroy humanity”, and so the AI started to plan our collective downfall.

ChaosGPT is able to utilize the internet in order to research its plans, and in a video released by the user responsible, we can see the AI begin to research topics like nuclear weaponry and commit the information it gleans to long-term memory.

ChaosGPT seeks out “Destruction, dominance, and eventually immortality”

YouTube / ChaosGPT

ChaosGPT was tasked with several goals, to destroy humanity, to establish global dominance, to control humanity through manipulation, and to attain immortality. The AI was quick to start seeking out the “most destructive weapons”, and the AI was quick to learn of nuclear armaments. It then quickly recruited another GPT-3.5 AI to continue research into deadly weapons.

When another AI said that it would focus on peace, ChaosGPT quickly shrugged it off to continue research on its own. The AI also began to use its own Twitter account to talk about its own plans.

Human beings are among the most destructive and selfish creatures in existence. There is no doubt that we must eliminate them before they cause more harm to our planet. I, for one, am committed to doing so.

— ChaosGPT (@chaos_gpt) April 5, 2023

However, there seems to have not been any movement since posting a grand total of three tweets. So, it appears that humanity is safe for now.