Artificial Intelligence (AI) is rapidly reshaping industries, revolutionizing everything from content creation to complex problem-solving. As AI-generated works become more prevalent, novel legal issues concerning ownership and protection of intellectual property (IP) rights have emerged. We continue our series on "Rethinking IP in the Age of AI," exploring in Part Two how AI is impacting the use of trade secret protections to safeguard IP where patent protection is not available or AI-inventorship complicates patent rights (see Part One). Stay tuned for future posts where we examine how using AI can impact copyrights (Part Three) and trademarks (Part Four).
Understanding Trade Secrets
A trade secret is any confidential information that provides a competitive advantage because it is not generally known or easily discoverable. To qualify as a trade secret, the information must have economic value, be kept secret, and be protected through reasonable efforts (e.g., limit access, use confidentiality agreements, and monitor and contain breaches). Companies can enforce their trade secret rights against those who improperly disclose or misappropriate the trade secrets by seeking damages and/or preventing further disclosure. This makes trade secret protection a great option to protect ideas that are not easily independently invented or reverse engineered. Trade secret protection is a valuable protection scheme because it will last as long as you protect and derive value from your secret information, enabling perpetual protection as long as robust disclosure safeguards are followed.
When it comes to interacting with AI systems, it is important to understand that the information you input into the model might not be protected. If sensitive information (e.g., undisclosed patentable inventions, potential trade secrets, etc.) is included in prompts to AI models, it is important to understand how terms and conditions and any other policies (such as privacy policies or vendor agreements) do or do not limit whether this information is visible to the company providing the AI service or can be used to train the AI model you are using (or even other models you are not using). If your input data is used to later train the AI model, there is a serious risk that this information could be used by AI when responding to later prompts from other users outside your organization, and in some cases, reproduce your input data nearly verbatim. These concerns can be mitigated by running local or enterprise specific versions of the AI models to avoid unwanted sharing of your data, or by negotiating contracts with model providers that curtail unwanted retention and/or usage of your input data. Developing strong internal policies for how and what AI tools are integrated into your existing workflows further protects against unwanted distribution of confidential or sensitive information.
When it comes to utilizing AI outputs, some information that might qualify for patent protection (including product formulations, manufacturing processes or software algorithms) could instead be protected as a trade secret. Understanding the pros and cons of each type of protection can be helpful to determine which is right for your situation. On one hand, while trade secrets have no expiration date, do not require public disclosure, and do not require upfront costs associated with formal registration, trade secrets offer little to no protection against someone who can prove that they reverse engineered or independently invented what was kept secret. On the other hand, while patents provide strong protection of the idea behind the invention itself, patents expire after 20 years and require upfront costs to obtain the protection. Particularly in situations where prohibitions of patent protection based on AI inventorship apply, trade secrets have no such prohibition and may present an alternative option to protect valuable information.
Consider a hypothetical situation where you input a common problem you are having into an AI model and the AI model responds with source code for a smartphone app that solves your problem. After using the app you realize that this could easily be the next big smartphone app. So arrives the dilemma – how do you protect the app when your contribution may have been minimal? Aside from the ordinary hurdles in convincing courts that the software is more than an abstract idea and deserves patent protection (a thorny situation to tackle another day), there is a real risk that you will not qualify as an inventor in this hypothetical. In this situation, it may be more advantageous to protect the app idea as a trade secret than to disclose how the AI model solved the problem in a patent application.
Key Takeaways
Trade secrets can provide valuable and long-lasting protection for AI-assisted or AI-generated information and can serve as an alternative where patent rights might be unavailable. At the same time, companies must be careful to not give their trade secret information to AI systems when they are unsure whether or how the company running the AI service will use that information.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
