Overview

Simon Willison tracked OpenAI’s legally-required IRS mission statements from 2016-2024, revealing how the company has systematically removed commitments to openness and community collaboration while strengthening language around control and deployment of AI technology.

View Original

The Breakdown

  • Tax filing analysis reveals mission drift - OpenAI’s IRS 501(c)(3) filings show legally-binding mission changes that reflect the company’s evolution from open research to controlled deployment
  • Abandonment of open collaboration - The 2018 filing removed language about ’trying to build AI as part of a larger community’ and ‘openly share our plans and capabilities along the way’
  • Shift from helping to controlling - Earlier versions wanted to ‘help the world build safe AI technology’ but later changed to ‘develop and responsibly deploy safe AI technology’ themselves
  • Confidence escalation in AI impact - Language evolved from AI will ‘most likely benefit humanity’ to definitively ‘benefits humanity’, while adding ‘safely’ as a qualifier in 2022
  • Terminology evolution reflects focus change - ‘Digital intelligence’ became ‘general-purpose artificial intelligence’, signaling a shift toward more ambitious and centralized AI development goals