Simon Willison: 2022 Altman Email Reveals OpenAI's Plans for an Open-Source Language Model

Main idea

A Sam Altman email to OpenAI's board from October 1, 2022 — exposed in the Musk v. Altman litigation — suggests OpenAI considered releasing a powerful open-source language model not for altruistic reasons, but primarily as a competitive weapon to preempt rivals and discourage investor funding for competing AI efforts.

Context

The Musk v. Altman trial concluded in May 2026, with the jury ruling for OpenAI on statute-of-limitations grounds. Extensive internal communications from 2022-2024 were disclosed during proceedings. Willison regularly documents transparency and accountability issues in the AI industry.

Why it matters

The email sheds new light on OpenAI's internal thinking about open-source strategy at a time when the company publicly cited safety as the primary reason for its closed approach. It suggests more complex motivations — relevant to ongoing debates about transparency among frontier AI labs.

Details / arguments

• Email dated October 1, 2022; addressed to OpenAI board
• Altman discussed releasing a GPT-3-class model for consumer hardware
• Intent: preempt rivals and discourage investors from funding competing AI labs
• Musk v. Altman: jury ruled for OpenAI in May 2026 (statute of limitations)
• Published July 20, 2026 on simonwillison.net