Ex-Employee's Warning
Zoe Hitzig, who recently departed from OpenAI, has voiced significant apprehension regarding the new advertising experiments being conducted with ChatGPT.
Her primary concern is that the advertisements displayed to free and 'Go' tier users, specifically excluding those under 18 or predicted to be underage, could leverage the incredibly personal details users confide in the chatbot. This includes sensitive topics such as health issues, personal crises, or intimate relationship details. Hitzig articulates that advertising built upon such an archive of intimate conversations presents a potent risk of user manipulation, creating scenarios that are currently beyond our capacity to comprehend or effectively counteract. The trust users place in AI for candid disclosures, often things they wouldn't share with other humans, makes the prospect of this data being used for targeted advertising particularly unsettling and intrusive, potentially leading to subtle, data-driven influence over user decisions and perceptions.
Privacy and Manipulation Risks
The core of Hitzig's worry revolves around the profound trust individuals place in AI chatbots, often confiding highly personal matters they might hesitate to discuss with anyone else. Using these deeply private exchanges as a foundation for targeted advertising is viewed as a significant privacy risk and potentially invasive. The fear is that such practices could pave the way for AI systems to subtly shape user behavior and opinions by exploiting the very vulnerabilities and sensitive information users have shared in confidence. This raises ethical questions about the boundaries of data usage in AI interactions and the potential for sophisticated manipulation that users may not even be aware of, underscoring the need for robust privacy safeguards in the development and deployment of AI technologies that handle sensitive personal data.
OpenAI's Reassurance
In response to these concerns, OpenAI has stated that the advertising tests are not designed to target users based on their personal data. The company indicates that initial ad placements are at the bottom of responses and are not intended to steer or influence ongoing conversations. Furthermore, OpenAI claims that advertisements are not specifically targeted at users under 18 or at sensitive topics that users might discuss. They assure that personal data is not shared with advertisers, with only general aggregate statistics being used. This stance aims to alleviate fears by emphasizing their commitment to user privacy and responsible ad implementation, asserting that the current testing phase prioritizes non-intrusive methods and data protection.
