OpenAI’s ‘Code Red’ Crisis Memo Teases New Model and Ads in ChatGPT

OpenAI CEO Sam Altman has declared “code red†at the company, refocusing all resources into improving the quality of its flagship product ChatGPT and delaying other initiatives, some of which are yet to be publicly announced, according to an internal memo from Monday obtained by the Wall Street Journal.

OpenAI has had a rough few months. The company garnered significant ethical and legal backlash after several cases showed that ChatGPT’s safety guardrails degraded in some user conversations, which might have led to serious and, at times, fatal mental health episodes for some users.

At the time, OpenAI executives took some measures to make the chatbot safer for users, like getting rid of some sycophantic language that users claimed gave the chatbot more personality, introducing parental controls, nudging users in marathon conversations, and preparing an age prediction system to automatically apply “age-appropriate settings†for minors.

But, according to a New York Times report from last month, OpenAI executives declared “code orange†in October (one step down from this week’s alleged code red), saying that ChatGPT was facing intense competitive pressure, that the safer chatbot was just not connecting with users, and outlining goals to increase daily active users for ChatGPT by 5% by the end of the year.

Also around that time, OpenAI announced that they would be relaxing some of the previous restrictions around chatbot safety by allowing the chatbots to have more “personality†and including “erotica for verified adults.â€

According to this week’s leaked memo, in the future, ChatGPT users can expect increased speed and reliability in answers, more personalization, and a chatbot that is capable of answering a wider range of questions. Head of ChatGPT Nick Turley echoed some parts of that statement in a post on X from Monday, saying one of the company’s areas of focus was to make the chatbot “feel even more intuitive and personal.â€

OpenAI is also reportedly planning to release a new reasoning model next week. In the memo, Altman claims that the incoming model beats the performance of Google’s latest release, Gemini 3, which received widespread praise and comparisons to ChatGPT when it launched last month. Many users, including Salesforce CEO Marc Benioff, have claimed that Gemini 3 outperforms ChatGPT by a considerable degree.

Ads might be coming to ChatGPT

Altman’s turnaround effort will allegedly include temporary team transfers and delaying other initiatives the company is working on.

Namely, OpenAI will be pushing back its work on other ideas like AI agents for health and shopping, and the personal assistant Pulse, all features that the company unveiled only a few months ago.

They will also be delaying work on advertising initiatives, the Journal reports. That initiative OpenAI is yet to acknowledge publicly, but according to The Information report, the company has been testing different types of ads in its chatbots. Even though Altman once described “ads-plus-AI†as “sort of uniquely unsettling,†company executives have admitted to toying with the idea of an ads business model for some time now, though with no initiatives ever officially announced.

Last week, engineer Tibor Blaho on X discovered that a beta version of ChatGPT’s Android app included several references to ads in its code, with implications that they will be used specifically in ChatGPT Search. In a post on X on Monday, Turley described Search as “one of the biggest areas of opportunity†at OpenAI.