OpenAI is experiencing one of its most tense moments since the launch of ChatGPT, to the point that the company is internally talking about a real "code red" due to the progress Google has made with Gemini 3.

Sam Altman has sent a memo to employees making it clear that the absolute priority now is to... to improve ChatGPT, even if it means slowing down other projects such as ads, AI agents for shopping or health, and the Pulse personal assistant.

The underlying message is crystal clear: OpenAI knows it no longer plays alone in the generative AI league and that its once unquestionable leadership has become, at the very least, debatable.

OpenAI's “code red” and the Gemini 3 blow

According to reports, Altman has described the situation as an internal “code network,” a kind of high-level alarm that forces teams to reorganize, prioritize resources, and meet daily just to discuss how to make ChatGPT feel a step ahead of the competition again. The decision comes right after the launch of Gemini 3, Google's model that is dominating benchmark rankings and which, in several public tests, has outperformed OpenAI's most advanced models.

It's not just a matter of technical metrics: Gemini 3 has been received as confirmation that Google, which for years seemed to be trailing behind OpenAI, has managed to catch up and even surpass it in some areas. This is already being reflected in traffic: some analyzes suggest that OpenAI has lost around 6% of its users since the launch of Gemini 3, a worrying sign in a market where inertia and the network effect are everything.

ChatGPT over ads and agents

The strategic shift is decisive: OpenAI has decided to pause or slow down key projects that just a few months ago seemed to be the future of its business. These include advertising initiatives within ChatGPT, AI agents for shopping and health, and the development of the Pulse personal assistant. Designed to be a kind of permanent “co-pilot” in the user's digital life.

Instead of aggressively diversifying,Altman has ordered a focus on the product that started it all: making ChatGPT faster, more reliable, and much more personal. Internally, the discussion centers on improving personalization, reducing slow or inconsistent responses, and refining the model's behavior so that it responds better to a wider range of questions without falling into unjustified blocks or overly conservative answers. There is also pressure to strengthen the reasoning and image generation capabilities, two areas where OpenAI wants to once again differentiate itself from Gemini 3.

Regaining a leadership that is no longer so clear

For much of the post-ChatGPT era, it was almost taken for granted in Silicon Valley that OpenAI was one or two steps ahead of Google in language models, both in capabilities and brand perception. Now the narrative has reversed: Gemini 3 is topping many performance charts, and Google is boasting, with data in hand, of outperforming GPT in most public benchmarks. It doesn't help that OpenAI has had recent issues with versions of GPT that users perceived as "colder," less useful for simple tasks, and overly hesitant to respond. For a company that isn't yet profitable and needs to justify multi-million dollar infrastructure investments, losing momentum right now would be a luxury it can't afford. Hence, this "code red" isn't just a technical response, but also a public relations move: OpenAI wants to send a signal to the market, users, and strategic partners that it will fight to reclaim its crown, even if it means shedding, at least temporarily, more ambitious dreams like becoming a major advertising platform or the invisible operating system for all its AI agents.

Ultimately, Altman's message is as simple as it is powerful: if ChatGPT ceases to be clearly the best chatbot, OpenAI's entire house of cards will collapse. For the first time since 2022, Google has managed to make OpenAI look towards Mountain View with genuine concern, not just respect, and that alone marks a turning point in the race for generative AI.