Sui Generis is attractive to developers – it gives them the ability to continue developing their invention without losing protection

Continued use of an innovation to maintain and extend protection is one possible avenue of determining duration. For example, in trademark law in the U.S., a declaration of use (or excusable nonuse) is required every ten years in order to maintain trademark protection.72 This provides owners an incentive to continue using the right (i.e. so that they can continue to maintain protection), while also allowing competitors or other innovators to create their own marks within the same field. Such a continued use duration, though, has downsides when applied to an AI. So many aspects of an AI invention, whether it in the input data, algorithms that comprise it, or output produced, can change throughout its “lifetime” that defining “use” to any single embodiment would be almost impossible, as that use could change often. In the trademark context, the mark (as well as the company to which it applies) will generally remain the same throughout the course of its use. A solution could be that each subsequent embodiment of an AI, developed over time, would be entitled to its own protection. However, such a scheme would require further administrative costs as each use would need registration (as outline above) as well as a determination of when the AI has changed enough to warrant further protection. Another alternative could be a set term of years, as is usually used in copyright and patent regimes. Given the speed at which this field develops, a shorter period may be desired, on the order of perhaps 5 to 10 years. This duration would depend upon a careful study of how these technologies develop as well as the costs related to that development and potential for recouping that cost in the market. Another alternative, used by the sui generis database right, is the option to renew the right for an additional term of years upon showing of “substantial investment.” This could be attractive to AI developers, as in this way they can continue developing their AI, changing the base algorithms, inputs, and outputs without fear of losing their protection. Of course, there is still the administrative cost of registering and tracking the AI changes, but in this case, they do not risk losing protection that they already have, but rather risk not gaining protection, if they fail to properly report and track their invention.