In a recent interview with The Atlantic, Sam Altman, CEO of OpenAI, discussed the idea of implementing a confidentiality agreement similar to attorney-client privilege for sensitive information shared with AI. Altman suggested that society may need to consider creating a concept of "AI privilege" in order to protect the privacy of individuals interacting with AI systems.
The conversation arose in the context of Altman and media mogul Arianna Huffington's new AI health venture, Thrive AI Health, which aims to provide personalized health recommendations based on user data. As the use of AI systems and products becomes more prevalent, the issue of regulating data storage and sharing has become increasingly important.
Altman highlighted the importance of ensuring that individuals feel comfortable sharing sensitive information with AI systems, particularly in the healthcare sector. He noted that some people have found it easier to open up to AI systems like large language models (LLMs) than to humans, which underscores the need for clear guidelines on data privacy and confidentiality.
However, concerns have been raised about the potential misuse of health information by big tech companies, particularly in the training of AI models without proper licensing agreements. Altman emphasized the importance of transparency and educating users about data privacy to address these issues.
Thrive AI Health aims to democratize access to expert-level health coaching and address health inequities through AI technology. While the specifics of the product are still being developed, Altman and Huffington are committed to ensuring that data privacy and confidentiality are prioritized in their AI health platform.
Overall, the discussion around regulating sensitive information shared with AI reflects the evolving landscape of AI technology and the need to balance innovation with privacy protection. As society grapples with these challenges, it will be important to consider how best to establish guidelines that promote trust and transparency in the use of AI systems.