2026-02-15

Added

The Evolution of OpenAI's Mission Statement

Type: Analysis / Investigation
Author: Simon Willison
URL: https://simonwillison.net/2026/Feb/13/openai-mission-statement/
Ref: openai-mission-evolution

Summary: Investigative analysis tracking how OpenAI's official mission statement evolved from 2016-2024 using their IRS tax filings (Form 990). The original 2016 mission emphasized being "unconstrained by a need to generate financial return," building AI "as part of a larger community," and wanting to "openly share our plans and capabilities." By 2024, the mission was drastically simplified to: "OpenAIs mission is to ensure that artificial general intelligence benefits all of humanity"—with notable omissions of safety language and financial independence.

Key changes: 2018 dropped openness/community language; 2021 shifted from "help the world build" to "develop and responsibly deploy" (taking ownership); 2024 removed all mentions of safety, financial independence, and open collaboration. Simon used Git to visualize the changes, creating a transparency tool that could be applied to other nonprofits' mission drift.

Why it matters: Provides legally-binding documentation (not just PR) of OpenAI's transformation from open, collaborative nonprofit to simplified AGI mission. The removal of "unconstrained by a need to generate financial return" aligns with their for-profit transition and multi-billion dollar valuation.

Related: openai, agi, ai-safety, anthropic, transparency-ai

Cognitive Debt in AI-Assisted Development

Type: Article / Commentary
Author: Simon Willison (on Margaret-Anne Storey's concept)
URL: https://simonwillison.net/2026/Feb/15/cognitive-debt/
Ref: cognitive-debt-ai

Summary: Cognitive debt is the mental burden and loss of understanding that accumulates when developers use AI to generate code without fully comprehending it. Unlike technical debt (which lives in code), cognitive debt lives in developers' brains and affects their ability to reason about and confidently modify systems. Margaret-Anne Storey describes a student team that "accumulated cognitive debt faster than technical debt" and became paralyzed, unable to make even simple changes. Simon Willison reflects on his own "vibe coding" experiments: "I no longer have a firm mental model of what they can do and how they work, which means each additional feature becomes harder to reason about."

Why it matters: Challenges the "move fast" approach enabled by AI agents. Even if AI generates clean code, developers who don't understand their own systems lose agency, face feature paralysis, and struggle with debugging. For teams, shared understanding evaporates when different members use AI to generate different parts of the system. This suggests code review and comprehension remain essential even with AI assistance - speed without understanding has a hidden cost.