ChatGPT, Generative Artificial Intelligence, and Trade Secrets: Generating Solutions or Generating Problems?

Download PDF of this page 

published 05/08/2023

The release of ChatGPT and other generative artificial intelligence (“generative AI”) has been a boon for many industries, from marketing to healthcare – or so it is perceived by most to be that way. The focus in the media has largely been on the beneficial uses of these systems in various commercial enterprises which include, but are not limited to, the following:

1. Software development and testing;
2. Research and development; and
3. Publishing, media, and writing.

Concomitant with the release of ChatGPT has been the encouragement of its use for the purposes of materializing or executing ideas (for example, telling the bot to write a 150-word story about a heroic butterfly conquering a dragon), but also for testing and developing ideas (for example, generating test cases for software development). The appeal lies in the fact that the bot can save a person time and effort; however, the indiscriminate use of generative AI can present real risks and be damaging to a business. Although the more common concern is copyright and ownership of the generated information, a particular risk that is often overlooked lies in the area of trade secrets.

Since AI such as ChatGPT learns from user input and generates output based on the input, care should be exercised by the user to not enter trade secret or proprietary information, as such information can inadvertently be reproduced by the AI and disclosed to third parties. In recent news, Samsung employees uploaded sensitive internal source code on ChatGPT, prompting the technology giant to issue a memo prohibiting the use of generative AI: Samsung Bans ChatGPT Among Employees After Sensitive Code Leak (forbes.com). One of the grounds for concern was the inability of Samsung to retrieve and delete the sensitive information that had been inputted into ChatGPT and stored in external servers. Another concern was the potential for the confidential information to be disclosed to other users. Other businesses, including Amazon, JPMorgan Chase & Co., and Goldman Sachs, have also placed similar restrictions on their employees in an effort to protect their valuable business information.

The increased use of generative AI has real implications for businesses. One might think that a workplace policy might be sufficient to protect trade secrets; however, businesses ought to further consider other circumstances in which generative AI might be used at the company’s risk. For example, a company’s employees might not be the only users of generative AI in producing work for the business. Contractors and other third parties might be using generative AI too. In this regard, a business ought to consider tailoring their contracts with service providers and contractors to ensure that the service provider or contractor does not use technologies such as ChatGPT that might disclose valuable information to others or make it difficult for the business to control and delete the information.

Co-author Dan Patriarca has left TDS to pursue a new opportunity, effective July 21, 2023. 

The Author(s)