# OpenAI CEO outlines three AI safety strategies for staff

> Sam Altman described three possible ways to keep advanced AI safe, including an IAEA‑style regulator, a merged‑lab model, and a worst‑case ‘single‑firm control’ scenario.

Oossa · 2026-10-20 · https://oossa.com/en/openai-ceo-outlines-three-ai-safety-strategies-for-staff

OpenAI chief Sam Altman told employees about three ideas for AI safety. The first mimics the International Atomic Energy Agency, proposing an independent body to audit models and monitor compute use. The second, called the “Demis Plan,” would combine major labs under one umbrella, similar to CERN. The third, the “Eliezer Plan,” warns that a single company could eventually dominate computing power and shut down rivals’ GPUs. Altman said the third is a theoretical extreme, but it highlights a risk some competitors worry about.

## The facts

- Altman presented the ideas last week, according to The Information.
- OpenAI, Anthropic, Google and Meta are working on a Standards Authority for Frontier AI.

## Why it matters

Understanding these proposals helps policymakers and industry players see how AI development could be governed or constrained, affecting future competition and safety standards.

## Sources & references

1. [Sam Altman reportedly pitched three endgames for AI safety, one involving a single firm melting everyone else's GPUs](https://the-decoder.com/sam-altman-reportedly-pitched-three-endgames-for-ai-safety-one-involving-a-single-firm-melting-everyone-elses-gpus/) – The Decoder, 2026-10-20

Last updated: 2026-10-20