Artificial intelligence can make inventions, hold convincing conversations and write poetry. What are the intellectual property implications?
AI systems are designed to mimic human problem-solving and decision-making capabilities, but at vastly higher rates. Recently, they have been designing magazine covers and identifying candidate drug molecules, for example. They learn by training themselves on large data pools, on which they hone their algorithms. They are adaptive and autonomous, and their workings can be mysterious. These characteristics have IP implications in terms of both access to training data and rights in AI output.
“Mining” and processing data can involve infringement of third party IP rights – in particular, copyright and, in the EU and UK, database right. Historically, copyright exceptions for data mining have only allowed non-commercial uses, and there were no applicable database right exceptions. The EU has now legislated for new exceptions where automated analysis is conducted to identify patterns, trends and correlations. It is still necessary to have lawful access to the material in the first place, however. Rights holders can also opt material out. Additionally, the EU Data Act will make it clear that data thrown out by machines (from jet engines to refrigerators) will not be protected by IP rights, and manufacturers must make it available to users and third parties (the latter subject to fair licence terms).
Playing catch-up, the UK ran a consultation on AI and IP that resulted in few concrete proposals for change, but will mean a new exception to copyright and database right for commercial data mining, with no opt-out. However, lawful access to the material (such as a digital subscription) will still be necessary. The proposal has stirred some controversy in publishing circles.
Another difficult, and uncertain, issue is that of rights in AI outputs. Repeated and, mostly, unsuccessful attempts have been made by a Dr Stephen Thaler to file patent applications citing the “DABUS” AI system as sole inventor. Courts in the UK and elsewhere have insisted that patents require a human inventor, whilst not ruling out that an AI system may have been used as an aid. A decision of the English Supreme Court is awaited. Many undecided points remain in this area apart from those in this case: for example, is it at all “inventive” to use an AI system to generate patentable subject-matter? And should the inventor and patent applicant respectively be the deviser, operator or owner of the system?
EU lawmakers seem to have concluded that AI-generated creative works should not be protected by copyright, largely on the basis that they lack originality in the legal sense. This is so even though the EU does protect photographs as copyright works based on perhaps minimal interventions (lighting, angle) by the photographer and whether or not they “pressed the button” themselves. The distinction seems to lie in whether or not the device is a mere aid to human creativity. In contrast to the EU, the UK and a handful of other territories grant copyright to works generated entirely by computer – in the UK, the copyright owner is the person (or company) making the “necessary arrangements” to create the work. It is not always going to be easy to identify that person when it comes to AI systems, however.
(Update, 3 February 2023: Given the perceived impact on the creative sector, the Minister of State told Parliament that “the proposals have clearly elicited a response that we did not hear when they were being drafted. We have taken the responses seriously… [W]e do not want to proceed with the original proposals. We will engage seriously, cross-party and with the industry, through the IPO, to ensure that we can, when needed, frame proposals that will command the support required.”)
This article is not legal advice, which it may be sensible to obtain before you take any decisions or actions in the areas covered. Please do contact me if you would like an initial discussion of your situation.