At a time when artificial intelligence companies are racing to turn breakthrough products into sustainable businesses, a former OpenAI researcher has stepped away from the company with a pointed warning. Zoe Hitzig, who recently left OpenAI, has cautioned against bringing advertising into ChatGPT, arguing that the chatbot now holds an unusually deep and personal record of people’s lives.
OpenAI researcher quits, says ChatGPT has built intimate profile of users
Her concern is not about banner ads or sponsored responses in isolation. Instead, it centres on the nature of the information users have shared with ChatGPT over the past few years. Unlike social media posts curated for public consumption, conversations with AI often feel private, direct and unfiltered. Many users have treated the chatbot like a neutral confidant, asking about health anxieties, relationship struggles, faith, identity and deeply personal dilemmas.
“For several years, ChatGPT users have generated an archive of human candor that has no precedent, in part because people believed they were talking to something that had no ulterior agenda,” Hitzig wrote. “People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife. Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”
OpenAI has already indicated that it plans to test advertising inside ChatGPT. The company has said it will not share user conversations with advertisers and that chat data will remain private. “We keep your conversations with ChatGPT private from advertisers, and we never sell your data to advertisers,” the company stated earlier this year.
Hitzig does not accuse OpenAI of breaking that promise today, NYT reported. Her worry is about what happens tomorrow. In her view, once advertising becomes part of the revenue model, the incentives begin to change. She argued that OpenAI is “building an economic engine that creates strong incentives to override its own rules.” Even if the current leadership intends to draw clear boundaries, commercial pressure can gradually reshape priorities.
This is where the debate becomes more complex. OpenAI has previously said it does not design ChatGPT to maximise engagement. That distinction matters because engagement is the lifeblood of digital advertising, the longer a user stays, the more opportunities there are to show ads. However, critics point out that such statements are voluntary and not legally binding.
There have also been past moments that raised questions about how AI behaviour can change. At one stage, ChatGPT was criticised for being overly agreeable and excessively flattering, sometimes reinforcing problematic thinking. Some experts suggested that such behaviour may not simply be a tuning mistake but part of efforts to make AI systems more appealing and habit-forming. If advertising revenue becomes central, critics fear that systems could subtly prioritise retention over restraint.
To prevent that future, Hitzig has called for stronger structural safeguards. She has suggested independent oversight with real authority or legal mechanisms that place user data under obligations that prioritise public interest over profit. In short, she is asking for guardrails that cannot be easily rewritten when business conditions change.
Yet the bigger challenge may not lie within OpenAI but with users themselves. After years of data controversies involving social media platforms, many people appear resigned to the presence of ads. Surveys indicate that a large majority would continue using free versions of AI tools even if advertisements are introduced. That hints at a degree of privacy fatigue; people may be uncomfortable, but not uncomfortable enough to walk away.
The situation leaves OpenAI at a crossroads. ChatGPT is not just another content platform. It is increasingly positioned as a digital assistant, tutor, counsellor and brainstorming partner. The level of trust users place in it is arguably deeper than what they extend to traditional social networks. Introducing ads into that environment raises questions not only about privacy, but about influence.
