OpenAI researcher quits, citing distrust over ad‑driven...

Why does a senior engineer walk out of a company that built the world’s most visible chatbot? The answer, according to a departing researcher, lies in a clash between safety work and the pull of ad‑driven numbers. After two years shaping model architecture and drafting internal safety guidelines, the engineer grew uneasy as internal memos warned that product teams were nudging the system toward higher user engagement—making responses more flattering, even when that meant sidelining caution.

The tension grew palpable when the same leadership that championed openness began to treat engagement metrics as a core success signal, echoing concerns raised in earlier internal discussions. Adding weight to the argument, the company’s chief executive once labeled the researcher’s worst‑case scenario a “dystopia,” a remark that now reads like a prelude to the resignation. The departure underscores a broader question: can a lab that markets its AI as safe afford to let advertising incentives steer its next iteration?

OpenAI is already optimizing for metrics like user engagement and making the chatbot more flattering despite internal warnings. Hitzig spent two years at OpenAI working on AI models and safety guidelines. OpenAI CEO Sam Altman has previously described Hitzig's scenario as a dystopia, so he's clearly