Artificial Intelligence: what kind of IP protection?



With this article we will give you a hint of IP and the latest trends in strategies for protecting AI.

1. Different types of IP protection

The best option for protecting AI is the adoption of a mixed-model approach, combining and taking the advantages of each type of IP protection applicable to AI. Here are the main types of IP protection of AI:

  • Copyright

    Artificial intelligence mainly consists of software—that is why it is also referred to as “super software”—which is typically protected by copyright.

However, copyright only covers the elements of expression of the creativity of the author. To this extent, copyright protection extends to the source code, which constitutes an expression of the creativity of the author, but not to the algorithm as such, which is considered a factual element.

  • Patent

    Programs for computers and mathematical methods as such are excluded from patentability; however, it is possible to patent the so-called “computer-implemented inventions” (CII), i.e. the inventions that have features realized wholly or in part by means of a computer program and which provide a technical effect.

    Therefore core-AI, e.g. the algorithm as such, is not patentable, whereas only AI-related inventions which have a technical character are eligible for patent protection.

    The European Patent Office has recently issued guidelines on the patentability of AI-related inventions, moving from the discussions carried out during the EPO conference on “Patenting Artificial Intelligence” held on May 30, 2018.

    When examining whether an AI-related invention is patentable, the EPO applies the two-hurdles approach:

    Moreover, when applying for a patent having an AI-related invention as subject matter, the following further requirements must be met:

    In this context, also the figure of the skilled person, who assesses the inventive step and the sufficient disclosure, is evolving and may be soon considered as a skilled human using AI.

    • 1st hurdle: The invention must have a technical character, i.e. it must have features which contribute to the solution of a technical problem by providing a technical effect.
    • 2nd hurdle: The technical features of the inventions shall involve an inventive step, i.e. they shall provide a contribution over the prior art which is not obvious to a person skilled in the art.
    • Clarity requirement: It is not always easy to use clear language when describing AI; EPO examiners recommend, inter alia, not to use marketing terms and buzzwords in the claims and description.
    • Sufficient disclosure: The patent shall be sufficiently detailed in order to allow a skilled person to be able to implement the protected AI invention by reading the claims and description of the patent.
  • Trade secrets
    Trade secrets grant an appropriate protection to the extent that the reverse engineering is not possible. However, protecting software form decompilation may be technically difficult and contractual restrictions thereto may be considered null and void according to the Software Directive (Directive 2009/24/EC). After all, it shall be considered that almost every IP right, and particularly patent protection, originates in a trade secret. In light of this, the best option is to combine trade secrets’ protection with copyright and patents. Ultimately, an open-source movement is also taking hold, encouraging open collaboration in developing AI.

2. IP ownership

In this context in which AI may create and invent autonomously, one of the main questions regarding IP protection of AI is, “Can computers and robots be considered as authors of AI creations?”

In relation to the copyright protection, this issue is a trend and different theories are being proposed, but pursuant to the European copyright system, the creativity shall be only that of a human being.

Similarly, within the patent system, the inventor should be a human being, but there is movement towards a change from an inventor-based system to an investment protection system.

Additionally, it is not easy to identify who is the author of a creation implemented with the use of AI between the creator of the AI system and the AI user.

In light of the above, it is fundamental to regulate the ownership of the IPRs by means of appropriate contractual clauses. To read more on how to regulate the ownership of the IPRs in IT service agreements, see also this article on AI and contracts.

3. What’s next?

The rapid evolution and spread of AI technology makes some legislative actions needed.

With respect to patent protection of AI-related inventions, in view of such rapid evolution of AI, there are suggestions to decrease the 20-year lifetime of patents and the 18 months secrecy period before the publication of a patent application.

Moreover, regarding the enforcement of patents covering AI, it has been noted that providing evidence of the infringement of a patent on an AI invention is difficult; hence, it has been suggested to consider reversing the burden of proof in this field.

Ultimately, one of the latest suggestions is to issue specific provisions to allow the patentability of core AI, i.e. algorithms as such.


DISCLAIMER: Because of the generality of this update, the information provided herein may not be applicable in all situations and should not be acted upon without specific legal advice based on particular situations.

© Dentons | Attorney Advertising

Written by:


Dentons on:

Reporters on Deadline

"My best business intelligence, in one easy email…"

Your first step to building a free, personalized, morning email brief covering pertinent authors and topics on JD Supra:
*By using the service, you signify your acceptance of JD Supra's Privacy Policy.
Custom Email Digest
- hide
- hide

This website uses cookies to improve user experience, track anonymous site usage, store authorization tokens and permit sharing on social media networks. By continuing to browse this website you accept the use of cookies. Click here to read more about how we use cookies.