Scuttlebutt

Scuttlebutt

"Trust is earned": OpenAI CFO on healthcare

7 takeaways: the months-long secret fix for a PHP flaw that could have "brought down a lot of the internet," and the model OpenAI held back after it solved a Millennium Math problem

Scuttlebutt's avatar
Scuttlebutt
Oct 23, 2026
∙ Paid

Masters of Scale · Oct 22, 2026 · 29m · 20m saved

With Sarah Friar, CFO of OpenAI

"Trust is earned": OpenAI CFO on healthcare

Photo: MoneyConf, CC BY 2.0 via Wikipedia

ChatGPT now has 1.2 billion weekly users — a fifth of the world's population — and OpenAI just crossed 100,000 advertisers across 89 countries. Sarah Friar, OpenAI's chief financial officer, told Reid Hoffman on his own show, Masters of Scale. The conversation was recorded live at the Masters of Scale Summit on October 21, 2026. The ad business exists, she said, so that roughly 90% of those users never have to pay for AI at all.

While rivals race to ship their most capable models as fast as possible, Sarah said OpenAI has deliberately done the opposite with its newest one.

"We've been working in a very secret way to get those vulnerabilities fixed before the offense discovers them." — Sarah Friar

As OpenAI's CFO, Sarah answers for both the revenue model and the security disclosures. On this stage, she is also the one who revealed that OpenAI spent months secretly fixing a vulnerability in the code underneath much of the internet.

The full interview is covered here so you can skip it. 29 minutes of audio, 9 minutes of reading.

Here are the 7 takeaways that matter.


This Scuttlebutt Summary is available in full to premium subscribers. If you're not one already, please consider subscribing to keep reading.


This Scuttlebutt Summary is available in full to premium subscribers. If you're not one already, please consider subscribing to keep reading.


Key Takeaways

  • OpenAI's ad business keeps ChatGPT free for about 90% of its 1.2 billion weekly users

  • OpenAI is holding back its most capable model — the one that solved the Navier-Stokes Millennium Math problem

    • The delay isn't "necessarily great for business in the short run," Sarah says

  • OpenAI secretly patched a PHP vulnerability that could have hit about 70% of websites

    • The fix took months, finishing before the flaw became public the same day as this interview

  • Healthcare is OpenAI's clearest proof that trust "is earned", not assumed

    • An HCA Healthcare pilot cut hospital length-of-stay by 25%

  • OpenAI's own staff use AI at 33x the rate of the most advanced outside firms

    • The gap between top AI-adopting companies and everyone else is now 8x, up from 2x in January

  • An $80 million community grant near a new Georgia data center funded free school lunches

    • Lunch-program uptake rose 50%

1. Ads Fund Free Access

Sarah opened by comparing the moment to the spread of electricity or the printing press, but said it is "so much more than that." She pointed to OpenAI's newly announced acquisition of Orchestra Bio, a platform pairing AI agents with scientists to speed up drug discovery, as evidence the shift is from AI that answers questions to AI that acts as "a virtual co-worker."

Ads keep ChatGPT free for about 90% of its 1.2 billion weekly users, Sarah says

"Our mission is AGI for the benefit of humanity, not AGI for the benefit of humanity who can pay for it." — Sarah Friar

ChatGPT now has 1.2 billion weekly users, a fifth of the world's population, and OpenAI has just crossed 100,000 advertisers across 89 countries — a business she said matters because it keeps AI from becoming pay-to-play.

2. Pacing The Frontier

User's avatar

Continue reading this post for free, courtesy of Scuttlebutt.

Or purchase a paid subscription.
© 2026 The Golden Age LLC · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture