AI creative engines can create new text, images, videos and audio. Is there intellectual property in this content and, if so, who owns it?
Generative artificial intelligence (AI) is capable of creating content in response to an input message. This may involve generating responses in text form in response to a question, or an image or video clip in response to a text message. Generative AI is a form of machine learning that has been trained on large amounts of data and has mapped patterns in the data, allowing it to generate similar data that is often very difficult to distinguish from human-created content.
Our previous articles have considered intellectual property (IP) issues in relation to input training data for generative AI: the risk of infringement, possible exceptions and possible reforms. But what about your results? Can they be protected by intellectual property rights and, if so, who owns them?
existing UK copyright law protects original literary, dramatic, musical and artistic works (LDMA), as well as films, sound recordings, broadcasts and published editions (so-called “business works”).
For LDMA works to be original, the work must be the intellectual creation of the author himself. Copyright protection for LDMA works lasts for the life of the author, plus an additional 70 years. When an LDMA work is created by a human with the help of an AI, as long as the work expresses original human creativity, the AI will be treated as a tool. The work will receive copyright protection like any other LDMA work and the rights will belong to the human author.
However, the UK also provides copyright protection for LDMA works generated entirely by computers, provided the originality threshold is met. Generative AI can put this originality threshold under pressure. Machine learning systems depend on their training data. A generative image-making AI system could create a digital image of an apple, but it would be shaped by the images of the apples in its training data. Whether this would undermine claims to originality has not been tested.
When the originality threshold is met, the “author” of such computer-generated LDMA work is considered to be “the person by whom the necessary arrangements for the creation of the work are made“. The protection lasts 50 years from the date of completion of the work, a figure notably shorter than the protection received when AI is used as a tool by a human creator.
Business works, unlike LDMA works, do not need to be original, but the rights are more limited and usually shorter, depending on the work in question. For example, sound recording protection only extends to a specific recording of a song and lasts for 70 years from its creation. These rights could apply to AI products. For example, if the AI generates a song and records it in the process, the person who took the necessary steps for the AI to generate and record the song is likely the producer and would therefore have the right to record the sound.
A recent consultation A study carried out by the UK Intellectual Property Office on AI and intellectual property considered the possibility of reforming copyright protection for computer-generated works. It concluded that no changes were necessary for the time being, but will monitor the adequacy of existing protections as AI develops further.
Unlike the UK, the EU does not offer copyright protection for creative works generated by AI. The Court of Justice of the European Union has not considered copyright protection of computer-generated works, but has confirmed that copyright protection requires some form of human intervention because it must reflect the personality of the author. Creative work could involve AI assistance, as long as the human input meets the originality/creativity requirement.
However, purely computer-generated works, by definition, lack any form of human contribution and, as such, are not eligible for copyright protection. A computer program may be protected by copyright, but any work you create autonomously would not be.
Products produced with AI may be protected by neighboring or related rights, such as sound recording rights (i.e. phonographic rights), film producers’ rights, etc. For example, phonographic rights enjoy rights of reproduction, distribution and communication to the public. Whenever he definition of a “phonogram” is fulfilled with an AI output, then phonographic rights could be available. When recordings are generated by AI, the right will be assigned to the producer of the phonogram. This is most likely the user of the AI software, as opposed to the AI developer, since the user will trigger the act of fixation (a necessary requirement to meet the definition of “phonogram”) when implementing the AI system. .
Patentability: the United Kingdom
In the United Kingdom, the Court of Appeal has held that an AI system cannot be an “inventor” for the purposes of filing a patent. The owner of an AI machine is also not entitled to a patent for an invention made by a machine when he claims that the machine was the inventor. This decision (part of the DABUS coordinated global litigation to test the limits of patent law regarding AI-generated inventions) has been appealed. The Supreme Court’s decision is expected before the end of September 2023.
This dispute has taken place in the context of the UKIPO. consultation on AI and intellectual property. Although no changes to UK patent law are currently planned, the government aims to boost discussions on harmonizing AI invention in international forums and develop a clear picture of what the UK should look for in any discussions. of harmonization.
However, human inventors are not prevented from patenting inventions where AI has been used as a tool, a point highlighted in the UKIPO consultation.
Similarly, in the EU, AI-generated inventions are not currently patentable. In a new iteration of the DABUS litigation, the European Patent Office (EPO) refused the notion that an AI system can be considered an inventor. Only humans can be named inventors in a patent application. An invention generated autonomously by an AI system without human participation will not be attributed to any inventor (not even the owner of the AI) and will not be patentable.
Again, inventions that use AI as a tool will be considered computer-implemented inventions. Although the European Patent Convention excludes computer programs “as such“Despite patent protection, inventions involving software can be patented as long as they are of a technical nature, that is, the invention produces a technical effect that serves a technical purpose. This could include providing a medical diagnosis using an automated system that processes physiological measurements, for example, using a neural network in a cardiac monitoring device to identify irregular heartbeats.
Autonomous or tool?
The fact that generative AI is trained to produce results that reflect the training data means that it could not be considered inherently novel or inventive; In fact, these systems have been called “pastiche machines” or “stochastic parrots.”
On the other hand, it is not uncommon for machine learning systems to generate results that humans had not anticipated. Something that at first glance seems “wrong” may turn out to be a new approach or idea that works. A famous example is the AlphaGo Zero AI system that defeated the human world champion using a move that no human had thought of in centuries of gaming. Therefore, it is not out of the question that, when asked how a particular process could be improved or undertaken, or asked to design a tool to address a defined problem, a generative AI system could generate content that can, if created by a human being, will be protected by intellectual property rights.
However, at the end of the day, the fact that a generative AI production does not occur without (human) input may influence it in favor of considering it a tool and not an autonomous creator or inventor.
Ownership and royalties for training data rights holders?
When licensing copyrighted works for training purposes, licensors could consider including terms that transfer ownership interests in the generative AI production and any associated intellectual property rights or seek royalty streams on any profits made from the productions. How this might work in practice remains to be seen. It is possible that the UKIPO will address this issue in due time. code of practices between rights holders and artificial intelligence companies on access to content protected by intellectual property.
Osborne Clarke comment
The development of generative AI is improving rapidly, which poses problems for the law to keep pace, especially in relation to the ability to protect the production of generative AI. As discussed above, copyright protection may be available in the UK but not in the EU.
Regarding the patentability of AI-generated inventions, the position in the UK (pending the Supreme Court’s DABUS ruling) and the EU is that it would be very difficult to obtain a patent unless it can be shown that the AI is a tool used by a human inventor.
The potential intellectual property protection granted to generative AI products is attracting attention beyond the UK and EU. For example, the United States Patent and Trademark Office recently issued a Request for comments on AI and inventiveness, focusing particularly on whether the law provides protection for AI-generated inventions.
There is concern about the extent to which creative industries will be affected by the increasing availability and speed of generative AI tools. Inevitably, a balance will need to be struck between the development of artificial intelligence technology and maintaining a robust intellectual property protection system that respects the interests of rights holders. The current UK government’s direction of travel is It’s still not clear.