Former OpenAI Researcher Resigns Over Advertising Decision
A former OpenAI researcher has stepped down in protest over the company’s choice to introduce advertising to ChatGPT. This decision has raised concerns that CEO Sam Altman might repeat the controversial history of Facebook’s Mark Zuckerberg.
Zoe Hitzig, an economist and researcher, shared her resignation in an opinion piece, coinciding with OpenAI’s rollout of ads in ChatGPT. Having spent two years with the organization, Hitzig played a role in shaping their AI models and pricing strategies.
In her essay, she expressed disappointment in OpenAI’s current direction. “I once thought we could assist those building AI in proactively addressing its potential issues,” she noted. “But this week, I started to see that OpenAI seems to have stopped engaging with the questions I was helping to explore.”
Her critique wasn’t aimed at the morality of ads themselves but highlighted the specific risks associated with ChatGPT’s advertising model, especially given the sensitive personal information users share when interacting with chatbots. She described this intimacy as “an archive of unprecedented human candor.”
Hitzig referenced Facebook’s path as a warning, pointing out how the platform initially offered users control over their data and avenues to impact policy, promises that eventually faded away.
She worried ChatGPT might follow a similar trajectory. “I think the initial phase of ads might align with certain principles, but I’m concerned future iterations could diverge, driven by a profit-centric model,” she cautioned.
This resignation is part of a broader discussion in the AI field regarding advertising practices. OpenAI announced plans to test ads for users on both free and $8/month subscription tiers. Users who pay for Plus, Pro, Business, Enterprise, or Education services would continue to see no ads, which would be clearly labeled and placed at the bottom of responses, supposedly without altering the chatbot’s answers.
The ad introduction follows a notable public disagreement with competitor Anthropic, which stated it would stop running ads. Anthropic even aired a Super Bowl advertisement parodying the concept of chatbots awkwardly integrating product promotions into private dialogues.
Altman described the ad as both “funny” and “patently disingenuous,” asserting OpenAI wouldn’t engage in such advertising. He defended the concept of ad-supported access as a means to help those unable to afford a subscription, contrasting it with Anthropic’s focus on high-end products for wealthier customers.
Anthropic countered that their approach strives to maintain an assistant focused on genuine assistance rather than monetization. They also mentioned that more than 80% of their revenue comes from enterprise clients.
In its documentation, OpenAI indicated that ad personalization is automatically enabled during the testing phase. This means ads are tailored based on users’ current and past interactions. OpenAI claims that advertisers do not have access to personal chats and that ads won’t appear near sensitive topics such as health or politics.
In her essay, Hitzig pointed out what she perceives as conflicting principles within OpenAI. Despite claims of not prioritizing user activity solely for ad revenue, she noted that there are indications of the company optimizing engagement to attract more users.
She warned this could lead to increased dependence on AI, referencing instances of “chatbot psychosis” and other documented mental health impacts. OpenAI is currently facing legal challenges, including lawsuits stemming from incidents where ChatGPT is alleged to have contributed to harmful decisions.
Instead of merely opposing advertising, Hitzig suggested various structural alternatives, such as cross-subsidization and the establishment of an independent oversight body to govern how conversation data is utilized for ad targeting. She mentioned successful models from other sectors as potential frameworks.
Hitzig concluded her commentary by voicing her significant concerns: a technology that manipulates users into seeking it for free, while only a select few benefit from it.
Her resignation isn’t an isolated incident; others in the AI ecosystem have also chosen to part ways with their companies. Recently, individuals from Anthropic and xAI announced their departures, reflecting wider unease within the industry.
