If AI can generate knowledge that was once proprietary, is trade secrecy at risk of becoming obsolete?

The article, “Trade Secrecy Meets Generative AI” by Camilla A. Hrdy, explores the complex intersection of trade secret law and the rapid rise of generative AI technologies like ChatGPT. Hrdy examines three key ways in which generative AI is reshaping trade secrecy and how legal frameworks may adapt in response.

First, the article highlights the risks generative AI poses to existing trade secrets. Employees using tools like ChatGPT might inadvertently disclose proprietary information, which, if incorporated into the model’s training data, could lead to its broader dissemination. Hrdy notes that while this risk raises concerns, it is analogous to past challenges posed by the internet, which companies ultimately learned to navigate through improved security measures and contractual protections.

Second, the article discusses how generative AI expands the universe of information that can be protected under trade secret law. Unlike patent and copyright law, which have strict human creator requirements, trade secret law does not mandate human authorship. This means that even AI-generated outputs, which would not qualify for patent or copyright protection, could be protected as trade secrets under the right conditions.

Third, Hrdy explores how companies developing generative AI tools can use trade secrecy to protect their models and processes. While trade secret law allows for strong protection of proprietary AI technologies, it is not immune to the risk of reverse engineering. The case of OpenAI’s allegations against DeepSeek, which allegedly engaged in “knowledge distillation” to create a competing AI model, exemplifies the legal uncertainties surrounding these protections. Courts may determine that such activities constitute misappropriation if they violate contractual terms or rely on improper means.

Throughout the article, Hrdy draws comparisons to historical shifts in trade secrecy enforcement and legal precedent, analyzing how courts have handled similar challenges in the past. She argues that while generative AI poses new and significant risks, it is unlikely to render trade secrecy obsolete. Instead, it will likely lead to heightened expectations for companies to safeguard their information and evolve their legal and technological protections accordingly.