OpenAI's Fidji Simo on ads in ChatGPT and ending the Code Red
How ads in ChatGPT will work, what will end the Code Red, those Anthropic attack ads, working with Sam Altman, and much more…
In her first extended interview since joining OpenAI last year, Fidji Simo, the company’s CEO of Applications, said that OpenAI’s internal “Code Red” will hopefully end with the release of a new model, that the company is exploring a social network where peoples’ AI agents interact on their behalf, and that ChatGPT’s ad model will look a lot more like Google’s intent-based system than the Facebook ads machine she helped build.
She also said ads will remain a minority of OpenAI’s revenue for “probably forever” — a claim worth stress-testing as the company mulls an IPO as soon as this year.
The full conversation was published today on a special episode of ACCESS, the podcast I co-host with Ellis Hamburger:
Simo’s role at OpenAI is unusual. Sam Altman oversees the research and compute teams. Everything else — ChatGPT, Sora, Codex, the enterprise business, ads, etc. — rolls up to Simo. “For me, as a person in charge of deploying, it’s like I’m a kid in a candy store having access to the best research lab in the world,” she said in the interview, which was recorded last Friday at her home in Los Angeles.
Highlights from the interview:
How ads will work in ChatGPT: According to Simo, the AI model itself won’t know that an ad even exists in a thread unless you ask about it. If you ask ChatGPT about an ad on your screen, she said it will give you a completely neutral assessment. “In some cases, the model is going to say, well, I see the ads for those two companies, but actually that one is good, but that one is actually not really good,” Simo said. She acknowledged some brands will hate this. “But I also think we live in a world where transparency and authenticity are what count.”
On the Code Red ending: OpenAI famously declared Code Red in December to accelerate ChatGPT’s growth after competitive pressure from Google and Anthropic. Simo told us the company is hoping to ship a new model soon that will end the Code Red. She made it clear that Code Red affected only the ChatGPT and product research teams and not the long-term research groups overseen by Altman.
An AI agent social network: When we asked about social networking ambitions beyond Sora, Simo went somewhere unexpected: “What happens to social relationships in a world where everybody has their own personal agents? And how can these personal agents help you manage your social relationships in a better way?” She framed it not as agents aimlessly talking to agents like Moltbook, but as AI mediating existing human relationships. “We haven’t cracked what that will look like, but I think that’s a very interesting area.”
The full episode covers a lot more, including Simo’s thoughts on AI companionship and loneliness, and what it’s like working with Altman.
Listen or watch wherever you get podcasts. Below is an edited transcript for Sources subscribers.
A MESSAGE FROM MY SPONSOR
A new model for health care
We’re working to prevent disease before it starts.
Too often, patients face barriers in getting the care they need. UnitedHealth Group is helping to remove these barriers while prioritizing new preventive care approaches that help keep patients healthy.
Transcript: Fidji Simo on ACCESS
The following conversation has been lightly edited for length and clarity:
