AI in Aviation: Is a Patent Owner Liable?

In part one of this two-part series, we outlined the legal issues at play as the aviation industry adopts AI in its operations. In this article, we cover another key question: To what extent is a patent owner liable for failures involving its AI-related IP?

A patent is a technical disclosure that protects an invention by granting the patent owner the right to exclude others from making, using, or selling the invention once it is patented. A patent does not protect the user from legal liability associated with the use or consequence of that invention. As a result, legal liability is more likely to occur from how the invention is used in real-world conditions. That said, poorly crafted patent applications can open patent owners up to liability. Patent owners can reduce their risk of legal exposure by doing four things in their patent applications:

1. Define the intended use of the AI technology in the disclosure of the patent: Clearly outlining the specific use and operating conditions in which the AI is configured to function can help establish the scope of responsibility for both the patent owner and the user. This may include a detailed description of the underlying technology and a clear explanation of the AI’s intended use.For example, consider a patent for an AI-based flight control system that is configured to assist in altitude adjustments during flight, but not configured for altitude adjustments during takeoff, taxiing, or emergency scenarios. The patent should explain these limitations and the system’s intended scope.

2. Document how the AI data is trained and under what conditions the AI data was gathered: Disclosing the training data and applicable conditions in which the data was gathered can help strengthen the technical credibility of the patent. This precise detail may help patent owners avoid any liability that may arise from performance of the AI in unexpected conditions.For example, a patent for the AI-based flight control system may be trained using data collected from normal weather conditions. Thus, the AI-based flight control system may not function properly under extreme weather conditions. Disclosing conditions in which the trained data was collected in the patent application can help patent owners avoid claims of misrepresentation or negligence if the system were to ever fail.

3. Describe limitations and boundaries of the AI technology: Explaining what the AI technology cannot do, its limitations, or where it is not intended to function can help set expectations and avoid the risk of overstating its capabilities. This careful approach may help a patent owner fend off claims based on misuse.For example, a patent for the same AI-based flight control system we discussed above may specify that the system is not intended to be used without human oversight.

4. Avoid absolute guarantees: To further reduce legal exposure, patent drafters should avoid using absolute terms in the patent application such as “critical,” “necessary,” “always,” and “must.” These terms can unnecessarily limit the scope of an invention in a patent application and may also lead to liability.

While patent protection is primarily a tool for granting the patent owner the right to exclude others from making, using, or selling the invention once it is patented, it can also serve as a tool in limiting any possible future liability.