OpenAI researcher quits over ChatGPT ads, warns of Facebook-style drift

A researcher at OpenAI has resigned in protest against the company’s decision to introduce advertisements in ChatGPT, warning that the move could erode core principles and expose users to subtle manipulation.

In a guest essay published in The New York Times, Zoe Hitzig said she stepped down on the same day OpenAI began testing ads inside the chatbot.

Hitzig, who spent two years at the company helping shape how its AI models were built and priced, said the shift convinced her she could no longer continue. “I once believed I could help the people building AI get ahead of the problems it would create,” she wrote, adding that OpenAI appeared to have stopped asking the hard questions she joined to help answer.

Drawing parallels with Facebook, Hitzig said the social network had once promised users control over their data and influence over policy decisions, only for those commitments to weaken as advertising revenue grew. She warned that while the first version of ChatGPT ads might follow stated principles, later iterations could override them as stronger financial incentives take hold.

Hitzig said ads themselves were not unethical and acknowledged that running an AI company is expensive, but expressed deep concern about advertising built on ChatGPT’s uniquely intimate data. She noted that users share sensitive details about health, relationships, and personal beliefs, and said advertising layered on such data could enable manipulation in ways that are not yet fully understood or preventable.

Concerns about ads in AI assistants have also been voiced by Demis Hassabis, chief executive of Google DeepMind, who said rushing ads into assistants could undermine user trust, even if advertising has funded much of the consumer internet.

OpenAI, valued at over $500 billion, has been seeking new revenue streams as it spends more than it earns. The company said ads will appear at the bottom of ChatGPT responses when a sponsored product or service is relevant to the conversation. Founded as a nonprofit, OpenAI last year restructured itself as a public benefit corporation.

Tags: