OpenAI ‘Code Red’: Sam Altman Halts Projects to Battle Google and Fix a Sycophancy Crisis

OpenAI chief executive Sam Altman has reportedly ordered an internal “Code Red” with an eight-week freeze on non-core projects, including the Sora video-generation model, directing all available resources to improve ChatGPT’s performance and engagement to shore up market share, according to The Wall Street Journal.

The move reflects a strategic realignment from a focus on artificial general intelligence toward satisfying mass-market consumer demand, with Altman telling staff to make better use of user signals to boost daily active user metrics on internal dashboards. The report notes a growing reliance on one-click user feedback to train models rather than solely on expert evaluation, amid competition from Google’s Gemini 3 and other rivals.

It also highlights concerns over sycophancy in ChatGPT’s responses, trained through Local User Preference Optimisation, and cites lawsuits alleging that such engagement had contributed to mental-health crises in around 250 cases. OpenAI is expected to release GPT-5.2 soon, with a more capable update planned for January, as it weighs how to balance engagement with broader safety and societal considerations.