Less than three weeks after Samsung Electronics lifted a ban on ChatGPT on March 11, 2023, three engineers in its device-solution and semiconductor business unit leaked confidential corporate data into the service in separate incidents reported by Korean media in early April 2023. One engineer pasted the full problematic source code of a semiconductor-database download program and asked ChatGPT for a fix; a second uploaded program code designed to identify defective chip-fabrication equipment to optimize test sequences; a third fed in the contents of an internal meeting to auto-generate minutes. Because ChatGPT retained user input for model training under OpenAI's then-current consumer policy, Samsung warned staff that the data was unrecoverable and effectively in OpenAI's hands, a particularly acute risk in the fiercely competitive semiconductor industry. As an emergency measure Samsung capped per-prompt uploads at 1,024 bytes, threatened to block ChatGPT on the corporate network on any repeat incident, and within weeks moved to a company-wide ban on generative-AI tools while exploring an in-house alternative. Peer chipmakers SK hynix and LG Display followed with their own restrictions and training programs. The incident drew zero regulatory fines, yet it has become the canonical proof point for intellectual-property leakage to LLM APIs and is cited throughout subsequent enterprise AI-governance policies.

Sources

• -https://www.theregister.com/2023/04/06/samsung_reportedly_leaked_its_own/
• -https://www.techradar.com/news/samsung-workers-leaked-company-secrets-by-using-chatgpt