A couple years ago, a powerful antibiotic, called Halicin, was discovered by a machine learning algorithm. 1 The algorithm had identified powerful new antibiotics from a pool of more than 100 million molecules. It was the first time an algorithm made a discovery from scratch without any previous human assumption or guidance. Today, we are using algorithms in many medical processes such as vaccine development, drug design, material discovery, etc. It is not a question of “if” but a question of “when” new inventions will be made by algorithms, and not only in the medical field. Algorithms have been creating and discovering new works also outside the medical field, for example in arts and music. As algorithms are now able to create and discover on their own, the question arises: Who or what is the owner?
Our current patent systems were written when we only knew humans as inventors. So, in the case where an algorithm creates something new, should the people who created the algorithm get the IP rights and acknowledgement for the algorithm’s contribution and invention? Or could the algorithm alone be credited as the inventor and hold the IP rights, or perhaps nobody gets the rights? In the case of the latter, new intellectual property laws and international treaties concerning copyright will be needed.
What or who does copyright protect?
Intellectual property (IP) law encompasses a range of legal rights that protect intellectual creations such as innovative works, inventions, software, etc. Copyright are the rights granted to creators of original authorship, like literary and artistic works. Copyright protection is automatically obtained when someone creates original literary, scientific, and artistic work. Nobody apart from the creator has the right to publish the work or reproduce it. The requirement imbedded in copyright laws states that there must be an 'own intellectual creation'. When you look up the definition of an intellectual creation, you find something along the lines of a "manifestation of a human expression of thought". As copyright protection laws were written for humans, not algorithms, do they still apply to works created by algorithms?
Currently, only humans can obtain copyrights
To answer the question whether algorithms can own the copyright to works it has created: no. The current copyright laws are written for humans, and only people can hold those rights. However, between 2011 and 2018, a series of legal disputes took place about the copyright status of selfies taken by a monkey using the camera of a British nature photographer. 2 The monkey had stolen the camera and took several photos with it, including a selfie that went viral.
Wikimedia hosted the images despite the photographers’ objections. The photographer claimed that he holds the copyright as he arranged the set-up that led to the pictures. His travel to Indonesia, befriending a group of monkeys, and setting up his camera equipment in a way that a "selfie" could be taken. On the other side, PETA, the People for the Ethical Treatment of Animals organisation, was arguing that the monkey should receive the copyrights. The judge ruled that neither the monkey nor the photographer can have the copyright under current law. According to the US Copyright Office, works "produced by nature, animals, or plants" cannot be granted copyright protection, and judges ruled there was no creativity involved from the photographer himself when the monkey took the selfie. The photographer’s arguments were hard to defend, which resulted in the photo’s entering the public domain.
In the situation where the monkey would have had an owner, the owner would also not receive any copyrights. It is most likely that the monkey did not even know that it was taking pictures, so it could not be considered as making creative choices. Therefore, no copyrights could be claimed even if the monkey had an owner.
Future copyrights for algorithms
Back to the algorithms, suppose that an algorithm can make truly original and creative choices autonomously, can the algorithm receive the copyrights?
With the fast paste of algorithms developing, their intelligence can soon no longer be distinguished from that of humans, there would be something to be said for giving algorithms copyrights and making them 'legal subjects'. For this to be realised though, new rules would have to be created. And what will happen to humans inventions, can they compete with algorithms?
Algorithms are changing our views on ownership, and this prompts not only a legal discussion, but also a philosophical one. Either way, IP professionals and lawmakers will have to deal with this.