Samsung is serving as an example of the challenges and pitfalls surrounding AI usage, with the tech company leaking sensitive info to ChatGPT.

Companies are grappling with how to integrate AI into their workflows, but doing so in a way that protects corporate secrets has proven a challenge. According to TheStreet, citing The Korea Times, Samsung leaked sensitive semiconductor data to the AI as engineers tested its capabilities.

One account of the leak states a Samsung staffer fed faulty semiconductor software data into ChatGPT to help solve a coding problem. Another employee fed company meeting conversations involving proprietary data to the AI chatbot to generate meeting minutes.

Such a scenario poses multiple problems, especially for a company like Samsung.

“Managers have also trained their workers on the scope of the use of this service,” The Times reported. “These actions are aimed at preventing data leaks by sharing internal information with the chat AI, given that semiconductor-related information is a state secret.”

What’s more, once data is fed into ChatGPT, there’s no way to undo it — the data permanently becomes part of ChatGPT’s knowledge base.

“The problem with sharing company secrets with ChatGPT is that those written queries don’t necessarily disappear when an employee shuts off their computer,” Gizmodo reports. “OpenAI says it may use data submitted to ChatGPT or other consumer services to improve its AI models.”

Companies do well to learn from Samsung’s mistake and set clear policies for what their employees can and cannot share with ChatGPT, Bard, and other AIs.