“Powerful AI systems could hold the key to some of the mega challenges facing humanity – from the cure for cancer to workable solutions for reversing climate change. But if outdated IP laws around the world don’t respond quickly to the rise of the inventive machine, the lack of incentive for AI developers could stand in the way of a new era of spectacular human endeavour.” Professor Ryan Abbot, University of Surrey, quoted in an article that appeared in The Patent Lawyer entitled First ever patent applications filed for inventions generated solely by AI.

In a previous article we have discussed artificial intelligence (AI) in the context of trade marks. In that article we looked at how AI, in the form of predictive devices like Alexa and Amazon Dash, has the potential to take humans out of the purchasing process, thereby raising interesting issues around trade mark infringement. In this article we discuss AI in the context of patents and copyright.

Patents

The abovementioned article in The Patent Lawyer discusses the fact that a team led by a university professor has successfully filed the first patent applications for inventions created by AI, with the inventor having been named as DABUS. DABUS is the name given to the machine in question and stands for “device for the autonomous bootstrapping of unified sentience”. Both the UK Intellectual Property Office (UKIPO) and the European Patent Office (EPO) have apparently indicated that there are no issues with novelty, inventive step or industrial applicability in respect of these inventions.

But what about the fact that the inventor is not human, but rather an AI machine or program? That’s the tricky part. According to the article there “have been claims of AI generating inventions for decades, but an AI inventor has never been disclosed in a patent application.” However, that is likely to change. “Modern AI may fundamentally change how research and development takes place. In some cases AI is no longer a tool, even a very sophisticated tool; in some cases, AI is automating innovation.”

The article explains that no country’s law specifically deals with the issue of whether or not AI inventions can be patented. However, most jurisdictions have restricted inventorship to natural persons, in order to prevent corporate inventorship. The article suggests that the right approach is for the AI to be listed as the inventor and for the AI’s owner to be recorded as the owner or assignee of the patent.

In another article dealing with the DABUS applications, a spokesman for the EPO is quoted as follows: “The current state of technological development suggests that, for the foreseeable future, AI is… a tool used by a human inventor.” We’re told that a UKIPO spokesman has also confirmed that an inventor must be a natural person. Whilst conceding that, change may well be required: “The government believes that AI technology could increase the UK’s GDP by 10% in the next decade, and the IPO is focussed on responding to the challenges that come with this growth”.

Copyright

Issues that were discussed in an article that first appeared in The WIPO Magazine in 2017, Artificial Intelligence and Copyright by Andres Guadamuz, have made news in South Africa of late.

There is provision in copyright law for computer-generated works, with the computer being seen as a tool employed by humans. But this, we’re told, is old hat – now machine-learning algorithms are learning from input data and generating new works through independent decision-making.

The author provides some mind-blowing examples: a novel written by a Japanese computer program that performed admirably in a national literary competition; a Google company called DeepMind that has created software that can generate music by listening to recordings; an AI program (funded by Google) that will write local news articles; computer programs that edit photos, write poems and write musicals – will any of us still have jobs?.

Perhaps the most astonishing of all is the so-called Next Rembrandt, the “Rembrandt” that was created by a 3D printer after analysing hundreds of Rembrandt paintings.

Guadamuz suggests that one approach is to argue that AI-created works can’t enjoy copyright because of the lack of human authorship, thus making AI-created works available to all. This may be an attractive approach for some, but it is likely to have a seriously chilling effect on the companies who sell the AI-created works – the music, the written works, the games... Companies will simply stop investing!

So, says Guadamuz, it is better to ensure that AI-created works do enjoy copyright, but who should own the copyright? This is where it gets difficult, because in many countries the law requires a human author. The USA, says Guadamuz, is such a country, as are Germany and Spain. Europe’s highest court, the Court of Justice of the European Union, has held that copyright requires originality and that originality must reflect the “author’s own intellectual creation.” It was decided in an Australian case that a work generated with the intervention of a computer had no copyright because it had not been produced by a human.

Guadamuz prefers the UK approach. UK legislation defines a computer-generated work as one that is “generated by a computer in circumstances such that there is no human author of the work.” It says that in the case of a computer-generated work the author is “the person by whom the arrangements necessary for the creation of the work are undertaken.” South African copyright law is substantially aligned with UK legislation in this regard when it comes to literary, dramatic, musical or artistic works, or computer programs, which are computer-generated.

But, says Guadamuz, the difficulty is that in the case of an AI-created work the computer program “is no longer a tool, it actually makes many of the decisions involved in the creative process without human intervention.” So who is it who made the arrangements for the creation of the work, the programmer of the program or the particular user of the program? This is where it gets complicated. Guadamuz suggests that the general principle should be that copyright belongs to the person who made the operation of the AI possible.

These are still early days and there’s certainly a long way to go. However, it is clear that the “IP in AI” ride is going to be an interesting one.