The capabilities of artificial intelligence (AI) continue to advance, with current systems able to write college-level essays, generate digital art from sentence prompts, and design technical solutions to complex problems.
The proliferation of AI systems has raised questions on the creation and ownership of various forms of intellectual property (IP). Join us for a program discussing the availability and ownership of IP rights when AI systems generate inventions and creative works.
Takeaways
Key Patent Considerations – AI as the Invention and the Inventor
- Software inventions, including AI inventions, can be protectable via patents, but the requirements of patentable subject matter must be met. In recent years, the United States Patent and Trademark Office (USPTO) has scrutinized software- and AI-related technologies, in particular under 35 USC § 101.
- Algorithms or computations that enable AI are not, on their own, patentable subject matter. They must be presented as being tied to a machine, such as a computer, with some improvement to the computer or functioning thereof and/or practical application or patentable concept.
- Novelty and nonobviousness are not provided solely by inclusion of AI, but AI models/datasets can support a case for novelty.
- In the United States and most other jurisdictions, an inventor has to be a human person. Computers (i.e., AI) cannot be inventors and shouldn’t be listed as such. Computers (AI) should be treated as a tool or contributor solely to reduction to practice. An inventor or inventors are defined as the people that conceived of the invention.
Key Copyright Considerations – AI as the Work, the Author, and the Brush
- AI code and implementations are protectable via copyright.
- The computer cannot be an author. An author is the person who does more with the output of the AI model. AI should be treated as a brush or medium.
Key Trade Secret Considerations – AI as the Informant or Confidant
- One can protect AI as a trade secret. For example, confidential AI models, frameworks, training datasets, etc., can all be trade secrets.
- AI-generated information can be a trade secret (but only if AI does not disclose the information outside of your organization).
- Trained AI generally operates as a blackbox. This is similar to object code vs. source code embodying a trade secret.
Key Trademark Considerations – AI as the Marketing Agency
- Registration of trademarks generally does not implicate AI. Generation of new trademarkable images can implicate copyright. AI cannot be an author of a registered copyright.
- AI trained on public datasets may inadvertently generate and use images or text similar to registered trademarks. AI can also intentionally create similar trademarks.
QUESTIONS?
For more information, please contact Amy Ramos.