IBM CTO Warns of Risks: ChatGPT May Leak Sensitive Data
IBM Automation’s Vice President and Chief Technology Officer, Jerry Cuomo, has raised concerns about the potential risks associated with using ChatGPT, OpenAI’s language model designed for enterprise use. In a recent blog post, Cuomo outlined several key areas businesses should consider before deploying ChatGPT.
Among the risks highlighted by Cuomo is the inability to control or understand how data is being used once it enters ChatGPT. This unintentional data leakage could pose legal consequences if customer or client data is inadvertently exposed to the public via ChatGPT’s training data. Cuomo warns that businesses may also face intellectual property risks and violations of open-source agreements due to data leakage.
IBM’s concerns are backed by concerns of unintentional data sharing. ChatGPT’s data model includes the possibility of sharing sensitive third-party or internal company information with individuals asking relevant questions. This raises concerns about data privacy and security for businesses relying on ChatGPT for their operations.
Cointelegraph reached out to OpenAI for a response to IBM’s claims. An intermediary responded, assuring that data will not be shared with individuals who ask relevant questions. OpenAI also directed attention to existing documentation on ChatGPT’s privacy features, including the ability for web users to disable chat history.
However, critics argue that while the ChatGPT API has data sharing turned off by default, conversations on the web version are saved by default. Users must actively opt out of saving their conversations and withholding their data from training the model. Until now, there is no option to retain conversations while opting out of data sharing.
It is evident that businesses using ChatGPT should be cautious about the potential risks associated with unintentional data leakage. While OpenAI has made efforts to address privacy concerns, there is room for improvement to ensure better control and transparency for users. As enterprises continue to embrace AI technologies, data privacy and security should remain a top priority.