ChatGPT ads could exploit users' personal info, warns ex-employee
Zoe Hitzig, who recently left OpenAI, is speaking out about ChatGPT's new ad tests.
She's worried that ads shown to ChatGPT Free and Go users (and not shown in accounts where the user is or is predicted to be under 18) could tap into the super-personal stuff people share with the chatbot—like medical worries or relationship struggles.
As she put it, "Advertising built on that archive creates a potential for manipulating users in ways we don't have the tools to understand, let alone prevent."
Hitzig is worried about how ads could affect users
Hitzig says people trust chatbots with things they might not tell anyone else, so using those conversations for targeted ads feels risky and kind of invasive.
She fears this could lead to AI systems subtly influencing people based on their own private info.
OpenAI says ads aren't targeting personal data
OpenAI says its ads are being tested at the bottom of answers to start and don't steer conversations.
They also claim they're not targeting users under 18 or sensitive topics, and personal data isn't shared with advertisers—just general stats.