This post was written by Barry Friedman.
Artificial Intelligence is proliferating quickly in the business environment. Employees are using ChatGPT and other engines to summarize documents, conversations, and prepare work product. While this is efficient and improves productivity, it can lead to serious unintended consequences.
Many AI platforms permit upload of documents for summary or as a basis for document generation. Additionally, users can input facts or other information for inclusion in output product. The important concern is whether this information or these documents contain confidential information – particularly confidential information of customers or other third parties.
The input of information to ChatGPT or other AI platforms is a publication. It is not confidential unless using a specialized platform with limited dissemination. Otherwise, each bit of information is incorporated into the language database that the AI platform uses to create product. This destroys any confidential status of the information and with respect to third party information, breaches any obligations of confidentiality.
Companies should have a developed policy for use of AI and take care to restrict use only to non-confidential information. If it is necessary to use AI for summary of confidential information or development of work product, it is important to ensure that the platform being used is limited in scope and use of the input data.
