Artificial Intelligence presents issues that require innovators to reconsider how to appropriately foster, protect, and enforce developments by their personnel. Patents can provide significant protection, but also present challenges in terms of patentability, inventorship, and novelty. Copyright can be used to protect software related innovations, particularly where a patent would not suffice or face significant obstacles, but can present authorship issues. Finally, trade secrets may provide protection where copyrights and patents cannot but can also suffer from drawbacks such as costly infrastructure investment to provide adequate protection. Once protected, an AI asset must then be leveraged, either internally or externally, to generate further value and justify the expense of its protection.

|

Creating AI Assets

Patents

Patents are increasingly being used to protect AI inventions. However, in view of precedent by the Supreme Court and the Federal Circuit, software patents can be challenging to obtain. Strategic patent drafting, though, can make the path to patentability more readily available. A pair of Federal Circuit decisions frame the issue.

In Electric Power Group v. Alstom, the Federal Circuit characterized the claims at issue as "collecting information, analyzing it, and displaying certain results of the collection and analysis." The claims did not assert any inventive step to transform them into patent-eligible subject matter, requiring no "new source or type of information, or new techniques for analyzing it." Rather, the court found that the claims relied on conventional techniques to perform well-known steps.

In contrast, the Federal Circuit's decision in McRO, Inc. v. Bandai Namco Games America Inc., provides an avenue for overcoming Electric Power type issues. There the Federal Circuit focused on claims reciting steps not previously employed by persons of ordinary skill in the art; claims directed towards generating animated facial expressions using a series of "morph weights." Ultimately, the court found the claims to be patent eligible because they reflected "a specific implementation" not shown to be known by other animators. Therefore, rather than attempting to obtain a patent on what results from an AI invention, claims should recite steps that an AI-related invention uses to achieve that result.

Copyright

AI programs themselves can be protected by copyright. But there is currently uncertainty regarding whether copyright protection can be extended to works created by an artificial intelligence. Even with this uncertainty, though, AI created works can still become a strategic asset when adequately protected, licensed, and used.

In the United States, only humans can be authors, as expressed by the Ninth Circuit in Naruto v. Slater, where the court dismissed a claim of copyright infringement brought by a macaque monkey. According to the court, Congress did not expressly grant standing to non-human animals under the Copyright Act, therefore only humans have standing to bring copyright actions. This does not mean that an AI created work cannot achieve copyright protection. Authorship may be attributable not to the AI program itself but rather to the creator of that program. However, the Copyright Office will refuse registration to any work created by a machine that is not accompanied by "sufficient human authorship" (Draft Compendium of U.S. Copyright Office Practices).

Relatedly, this begs the question: what happens when AI infringes on other copyrighted works? The question here turns on whether the user created access to the copyrighted original. Where the user has had little to no involvement, they are likely to face little liability, since they did not "provide access" to the copyrighted original. However, where AI users define even a limited data set that includes the copyrighted work, there may be an argument that the AI user created access.

Trade Secret

Trade secret law provides particular benefit for AI innovations because it allows for protection of two important elements: algorithms used by the AI and specific data used by an AI. In order to take advantage of trade secrets, a company must ensure appropriate privacy infrastructure is in place. This infrastructure can be costly, and should be weighed against the actual benefit gained from the underlying data/algorithm. For example, where a business has merely collected readily available data, investment in privacy infrastructure may not be worthwhile. But if a company has invested heavily in a proprietary algorithm it can be worthwhile to invest in significant privacy infrastructure.

Additionally, given the increasing prevalence of data privacy laws throughout the world, like the GDPR and CCPA, collected data will continue to increase in value, making trade secret law all the more important in ensuring that collected data stays valuable. As such, proper investment in privacy infrastructure can protect value. But in order to generate value, these AI innovations must also be properly leveraged as part of a firm's business strategy.

|

Leveraging AI Assets

Internal Use

One advantage that AI assets provide to businesses is the automation of basic, repetitive tasks. Rather than viewing AI as a "job-stealer," it should be viewed as an opportunity to free employees from repetitive tasks in order to focus on more creative thinking and problem solving.

AI can be used to increase the speed and accuracy of research projects and allow greater interaction between different projects, resulting in more collaboration than previously possible. AI assets can also be used to analyze the business as a whole by comparing different data from different parts of the company that before would not have be considered together. Using this data, an AI asset could propose new business ventures, determine where inefficiencies lie within the company, and ensure that repetitive work is curtailed.

External Use

AI can also be used to create more fruitful interactions with customers. Studies have shown that firms which incorporate AI assets into their customer interactions see an increase in both sales and inbound customers. By observing customer data and behavior, AI can allow for better prediction of customer dissatisfaction, as well as predicting improved customer solutions. This analysis can further help predict which proposed ventures will be worth investment.

One final way in which AI assets can be utilized externally is by providing a means of IP asset monitoring, allowing a business to identify low-value IP that it might not otherwise take notice of. Another use is tracking potential acts of patent infringement. Several startups utilize AI to monitor infringement of various types of IP, both physical and ephemeral. Such tools can simplify the monitoring of IP assets and increase the accuracy of finding infringing activity.

In the end, different types of AI can require different IP protection, dependent on both the form of the AI innovation and the applicable business strategy. The ultimate use for any AI innovation should be considered in connection with the form of protection desired in order to maximize both value and efficiency.

Mr. Ragusa a partner in the New York office of Baker Botts LLP, where his practice encompasses all aspects of intellectual property law including the prosecution, licensing and litigation of patents relating to artificial intelligence. The author appreciates the valuable assistance provided by Nick Palmieri in the preparation of this article.