Policy & Ethicsllmprivacy concernshallucinationsproprietary data
ChatGPT Presents Privacy And Safety Risks
8.0
A technology commentary warns that ChatGPT, now widely used since its 2022 release, is unsuitable for personal, illegal, proprietary, or medical queries because it collects prompts, can hallucinate, and has been involved in data exposures. The article cites a Cornell study showing models can reproduce training data verbatim and urges users to avoid sharing sensitive information or relying on the chatbot for legal or medical decisions.
Key Points
- 1Warns that ChatGPT collects and stores user prompts, making chats non-private
- 2Highlights model hallucinations and Cornell study showing near-verbatim reproduction of training data
- 3Advises practitioners to avoid inputting personal, proprietary, illegal, or medical information into chatbots
Scoring Rationale
Practical privacy and safety guidance raises awareness, but offers limited new evidence beyond existing concerns.
Sources
Public references used for this report.
Practice with real Logistics & Shipping data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
High-Value Overnight OrdersEasyDelivered International ShipmentsMediumOn-Time Delivery Rate by CarrierHard
250 free problems · No credit card
See all Logistics & Shipping problems
