The dramatic rise in artificial intelligence systems and software has led to an explosion in AI-generated information. Much of that information has commercial value, whether it be in the form of creative content or elaborate machine-learning algorithms and their resulting data streams. But how should that valuable information best be protected?
Traditional intellectual property regimes have been relatively inhospitable to AI-generated content. In the patent world, the Supreme Court's seminal decision in Alice Corp. v. CLS Bank held that a mere instruction to a computer to implement an abstract idea (for example, an algorithm or prompt) could not be patented.¹ And the U.S. Copyright Office recently issued guidance confirming that AI-generated content without any element of human creative contribution is not copyrightable.²
Fortunately, federal and state trade secrecy laws provide a ready alternative for valuable AI-generated information by expressly recognizing that trade secret rights "protect items which would not be proper subjects for consideration for patent [or copyright] protection."³ All that is required is that information be competitively valuable and secret. No application, registration, or disclosure is needed and there is no statutory expiration date for a trade secret. Furthermore, there is no requirement that the creator of a trade secret be a natural person (i.e., a human), as is required to secure a patent.
All of this is welcome news to the many companies founded on the promise of machine learning systems. Their algorithmic inputs, training data, neural network designs and implementation, and systems output data all must be protected as trade secrets if they are to be protected at all. But, in practice, maintaining secrecy can be challenging for companies that must partner with other business organizations or release their technology into the wild. Indeed, lawful reverse engineering as well as independent development can provide viable defenses to a trade secret misappropriation lawsuit. Therefore, companies would be wise to implement the following steps to protect their AI-generated data as trade secrets:
- Ensure non-disclosure agreements are entered into with all partner companies, independent contractors, and employees/officers/directors;
- When releasing software, ensure that it is designed and implemented in ways that cannot be reverse engineered; and
- Maintain strict cyber-security and monitoring protocols that are designed to prevent and detect data theft.
By taking these steps, companies will be in a stronger position to bring a lawsuit, obtain a restraining order, and collect damages if a trade secret is misappropriated.
Footnotes
1. Alice Corp. v. CLS Bank Int'l, 134 S.Ct.
2347, 2355 (2014).
2. Copyright Registration Guidance: Works Containing Material
Generated By Artificial Intelligence, U.S. Copyright Office (Mar.
16, 2023), available at
https://www.federalregister.gov/documents/2023/03/16/2023-05321/copyright-registration-guidance-works-containing-material-generated-by-artificial-intelligence
3. Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 482-83
(1974).
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.