OpenAI has begun testing ads on ChatGPT, prompting former researcher Zoë Hitzig to resign and publicly criticise the company’s direction. Writing in an opinion essay, Hitzig said the move risks repeating the same mistakes Facebook made with its advertising-driven model.
Hitzig said she does not consider advertising inherently unethical. She acknowledged that AI systems require expensive operational costs because they need continuous revenue to function properly. However, she warned that ChatGPT users have shared deeply personal information under the belief they were interacting with a neutral system.
According to Hitzig, building targeted advertising on top of that archive creates risks of subtle manipulation. She maintained her position that engagement anorganisationsed incentives would eventually diminish internal security measures for organizations because of their similar impact on Facebook security systems.
Hitzig established acontrol,tion between Facebook's initial commitments to user data management and user control which later collapsed under advertising demands. She suggested similar risks could emerge if OpenAI prioritises growth metrics such as daily active users. Reports have indicated the company optimises for engagement, which critics say may encourage overly flattering or dependent user interactions.
Hitzig proposed alternatives, including cross-subsidies where enterprise AI profits help fund broad public access. She also called for binding oversight structures with independent representation over how conversational data is used.