IBM Blockchain expert Jerry Cuomo has cautioned about the potential risks associated with utilizing ChatGPT for enterprise purposes.
In a recent blog post, Cuomo highlighted risk areas where businesses should be cautious before integrating ChatGPT into their operations. He highlighted that only non-sensitive data should be entrusted to the platform due to concerns surrounding data control and utilization.
Cuomo is worried that important information might accidentally leak out. This could be a problem if private partner, customer, or client details are shared in ChatGPT’s training data and become public. He’s also worried about intellectual property and breaking agreements about sharing information.
The potential for confidential third-party or internal company information to be incorporated into ChatGPT’s data model and shared with others who pose relevant queries is highlighted by Cuomo.
OpenAI, the producer of the model, has been constantly clarifying it’s policies on user protection and privacy. CEO Sam Altman even claims: “openai never trains on anything ever submitted to the api or uses that data to improve our models in any way.”
However, Cuomo points out the existing documentation on ChatGPT’s privacy features references, along with the option for web users to disable chat history. He believes it might not be ideal for enterprises to handle security and data leakage, IP issues, confidentiality and liability, open-source licenses, limitations on AI development, and uncertain compliance with international laws.
Will ChatGPT reach it’s ideal success?
By default, the ChatGPT API doesn’t let data be shared, but some people are concerned because conversations on the website version are saved automatically.
People need to say no if they don’t want their conversations and data to be shared; there’s no way to save conversations without agreeing to share data.