By Michael Scott
Jerry Cuomo lists several areas of concern, highlighting the likely dangers for firms utilizing ChatGPT.
Recently, Jerry Cuomo, IBM Automation’s chief technology officer, wrote a blog depicting what he believes are the risks linked to using ChatGPT for the enterprise.
The blog post indicates that businesses must consider several key risk areas before using ChatGPT. He also added that only non-sensitive information is safe with this chatbot. In this case, one lacks knowledge or control over the use of data that enters ChatGPT.
ChatGPT Vulnerable to Unintentional Data Leakage
The post also indicates that from a legal perspective, this form of unintended data leakage could jeopardize firms. This can happen when client or partner information is revealed to the general public following its leakage into the chatbot’s training data.
Jerry also mentions intellectual property (IP) risks and the likelihood of the leakage putting firms in open-source agreements contravention.
The IBM blog post shows that if the internal company or sensitive third-party information finds its way into ChatGPT, it becomes part of its data model and is shareable with persons who ask appropriate queries.
Chatbot Privacy Features Captured in Product Documentation
A public relations intermediary responded through email, claiming the data cannot be disclosed to persons who ask appropriate questions. Further, the representative cited the existing documentation concerning the chatbot’s privacy features, which includes a blog post explaining the web users’ capability to deactivate their chat history.
OpenAI claims that by default, the ChatGPT has data sharing deactivated. Nevertheless, critics argue that the web version chats are saved by default. Besides, users should withdraw from saving their chats and utilizing their information to train the model. The conversations feature is crucial to picking up where one left off. The choice to hold chats without approval to share information is nonexistent.
The concerns for the enterprise version arises from concerns expressed earlier this year by Italian, German, and French authorities over data leak. While OpenAI confirmed resolving the issues, tech experts are concerned over potential breaches by cybercrime experts.