In April 2023, Samsung's DS (Device Solutions) division discovered three separate incidents where engineers had pasted sensitive internal data into ChatGPT: one leaked semiconductor source code asking ChatGPT to debug it, another uploaded confidential meeting minutes asking for a summary, a third leaked proprietary test patterns. Samsung couldn't retrieve the data — once submitted to OpenAI, it was assumed to be training data. The company banned generative AI tools for all 300,000 employees and began building an internal clone.