The Great Mission Deletion: What OpenAI’s Evolving Statement Reveals

Simon Willison recently dug through OpenAI’s IRS filings and reconstructed how their mission statement has evolved from 2016 to 2024. It’s a fascinating archaeological dig into the tension between AI safety, open source values, and commercial reality.

The Core Insight

The changes aren’t random—they tell a story of an organization gradually walking away from its original commitments while maintaining the legal language of nonprofit status. The 2024 mission (“ensure that artificial general intelligence benefits all of humanity”) is notably shorter than the 2016 version, and notably missing any mention of safety.

Why This Matters

2016: The Idealistic Beginning
OpenAI’s original mission emphasized:
– Benefiting “humanity as a whole”
– Being “unconstrained by a need to generate financial return”
– Building AI “as part of a larger community”
– Openly sharing plans and capabilities

This was explicitly anti-commercial. They weren’t building a product—they were building a movement.

2018: First Retreat
They removed the commitment to “build AI as part of a larger community” and sharing openly. The financial return constraint remained, but the openness was gone.

2020-2022: The Gradual Erosion
– “Digital intelligence” became “general-purpose artificial intelligence”
– “Most likely to benefit” became simply “benefits”
– They shifted from “helping the world build safe AI” to “deploying safe AI” themselves
– “Safely” was added, then…

2024: The Deletion
The entire mission was reduced to 22 words: “OpenAIs mission is to ensure that artificial general intelligence benefits all of humanity.”

No safety. No openness. No financial constraints acknowledged. Just “benefits all of humanity”—which is conveniently vague enough to justify almost anything.

Key Takeaways

• Original mission emphasized openness and community building
• 2018 was the first major retreat from openness
• 2020-2022 saw gradual shift from “helping the world” to doing it themselves
• 2024 removed safety language entirely while keeping nonprofit status
• The tension between nonprofit structure and commercial ambitions remains unresolved

Looking Ahead

OpenAI isn’t alone in this evolution. The entire AI industry is grappling with how to balance safety commitments with commercial pressure. What’s notable is that they’ve managed to keep their nonprofit status while behaving like a commercial company—legally, they’re still bound by that 2016 mission, even if it’s been hollowed out.

The question isn’t whether OpenAI changed—it’s whether the industry has any mechanism to hold AI companies accountable to their stated missions when those missions are increasingly decorative.

Based on analysis of Simon Willison’s IRS filings research