It seems to be 100% mandatory if you fail some heuristic, which doesn’t happen that often, otherwise it’s only required for o3, o3-pro, reasoning summaries, and the new image generation.
I made a thread about it: OpenAI - We need to talk about org verification
The problem is, once a company gets big enough, it stops trying. Once you gain full dominance and saturate your market to where there’s no room to grow anymore, there’s no longer any need to appease anyone besides your investors. That’s when corners get cut, reputation becomes unimportant, and the focus shifts from rapid growth over to rapid profits.
Now that companies and enterprises have integrated OpenAI deeply within their code and average internet users are all reliant on - or even addicted to ChatGPT, there’s no chance that they’re losing their market share at this rate. As long as they roughly keep up with some direct competitors, they’re fine. So for us, it’s only downhill from here.
Remember, it’s unconstitutional for the Army to quarter AI troops inside your home. 