Researchers working on the AI Safety Camp 2026 project have published a detailed framework for mapping existential risk scenarios. According to the team, the model is structured as a decision tree dividing potential catastrophes into two main pathways: non-AI risks such as nuclear conflict and pandemic disease, and AI-driven risks, which further split into scenarios with multiple competing AI systems versus a single dominant superintelligent system. Each branch is mutually exclusive—probability assigned to one pathway necessarily reduces the likelihood of others. The framework aims to establish a shared taxonomy for ongoing debates about AI and existential danger.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton