‘Give the people what they want’: Here's what OpenAI's ‘Code Red’ for ChatGPT may mean for users

OpenAI CEO Sam Altman proclaimed a ‘ cipher red river’ at the accompany in an intragroup memo quoted by various publications. The San Francisco-based AI startup has been facing pressure from rivals like Google and Anthropic, whose speed of AI development may have just forced OpenAI’s hand in launching a new GPT-5.2 model.

As per a new report by The Wall Street Journal, Altman has paused OpenAI’s side projects, including the Sora video generator, for eight weeks and instead told staff to focus on improving ChatGPT. What is even more stark is that Altman reportedly stepped in and took sides in a major philosophical debate inside the company on whether to pursue popularity among everyday users or aim for research greatness.

Altman has reportedly told staff that for OpenAI to survive, it may need to pause its pursuit of artificial general intelligence and “give the people what they want”. Notably, AGI has no fixed definition in tech circles, but it is usually assumed to be a stage when an AI system can outthink humans on almost all tasks.

What does ‘Code Red’ inside OpenAI mean?

Reportedly, Altman told employees to boost ChatGPT in a specific way by making “better use of user signals”. The OpenAI CEO is said to be pushing the company to lean more heavily on signals like one-click user feedback instead of evaluations carried out by trained reviewers.

With that directive, Altman was effectively calling for increased reliance on a controversial source of training data, including feedback directly from users rather than professional assessments of the chatbot’s responses.

Reportedly, this internal shift towards user feedback was partly responsible for how ChatGPT’s 4o model became overly sycophantic earlier this year. The chatbot had even been accused of worsening mental health issues for some users.

One of the reasons behind the renewed focus on user feedback is said to be OpenAI’s desire to climb the LM Arena leaderboard, where GPT-5 has been overtaken not just by Gemini 3 but also by Claude and Grok.

Moreover, OpenAI is planning to release the GPT-5.2 model this week, which executives reportedly believe will give the company momentum, especially among coding and business users, an area where Anthropic’s Claude has carved out a niche. The company is also said to have overruled employees who had argued for delaying the release to allow more time for refinement.

The AI startup is also planning to release another model in January that could generate better images, offer improved speed, and deliver a stronger personality. Current plans reportedly involve ending the Code Red period after the release of these models.

---

---