NDAs have long been a cornerstone of protecting confidential data. In fact, in some circumstances it is the only legal document that will protect ideas and proprietary information. The use of AI in everyday tasks such as data analysis, however, introduces new risks and considerations that traditional NDA’s often overlook and addressing the use of AI becomes crucial to the protection and use of your data and information.
Here are five reasons why you need to address AI in Non-Disclosure Agreements:
- AI Systems Can Retain and Learn from Data. Despite the headlines, AI isn’t human (yet), and unlike humans, AI doesn’t forget. AI tools often process and store data in ways that are difficult to fully erase. If confidential information is fed into an AI model, it may become part of the system’s training data, creating a risk of inadvertent disclosure or misuse later. Agreements should explicitly prohibit using confidential data for training or improving AI models.
- Third-Party AI Tools Increase Exposure. Many organizations rely on external AI platforms for analytics, automation, or content generation. Without clear contractual language, confidential information could be shared with these third parties, violating the spirit of the NDA. AI clauses should require prior consent before using any external AI service with sensitive data.
- Regulatory Compliance and Liability. Data privacy laws impose strict rules on how personal and sensitive data is processed. To add to that burden, there are over twenty state data privacy laws, not to mention the spattering of AI related laws that address certain aspects of AI (but definitely not all). If AI systems handle confidential information improperly, both parties could face legal consequences.
- Defining “Use” in the Age of AI. Traditional NDAs focus on human actions, but AI introduces automated decision-making and data analysis. Provisions should clarify what constitutes “use” of confidential information by AI systems, including storage, processing, and derivative outputs, as well as who owns all of the resulting outputs.
- Future-Proofing Your Agreements. There may be no real “future proofing” but including AI-specific language helps your NDA remain relevant as new tools and capabilities emerge.
Here are some examples of provisions that you need to consider:
First, the nature of an NDA is of course, to share confidential information. Make sure your NDA addresses the prohibition of using your confidential information for AI training or to fine tuning. Second, ensure you have consent to share your data (and perhaps that of your clients or customers) with any external AI tool. Third, specify encryption, deletion and audit requirements. And finally, all parties must adhere to all applicable data protection laws.
For more detailed information or to schedule a consultation, please contact Ashley Brooks at ABrooks@RothJackson.com.
