ChatGPT privacy risks and functional limitations highlighted

As AI chatbots like ChatGPT become integrated into daily work and study, experts are warning users about significant privacy risks. Jennifer King from the Stanford Institute for Human-Centered AI notes that users often share more information than necessary because they do not perceive the chatbot as a corporate entity. However, once data is submitted, users lose control over how it is stored or utilized.

Security concerns are highlighted by past incidents, such as a 2023 OpenAI technical error that allowed some users to view others' personal information, including names, emails, and payment details. Experts advise against sharing personally identifiable information, such as ID numbers, passports, home addresses, or full dates of birth, with AI platforms.

In terms of functionality, recent testing shows that while ChatGPT can replace several smartphone applications for tasks like calculations, weather forecasting, and information retrieval, it still faces limitations. It struggles with tasks requiring real-time hardware integration, such as functioning as a reliable alarm clock or a stopwatch, suggesting that specialized applications remain necessary for deep operating system integration.