For five years, the Dabus question has dogged almost every UK court instance. Yesterday, the case reached the Supreme Court, whose judges must finally decide whether an AI system can be named as an inventor. While the verdict is unlikely to detract from previous decisions to the contrary, it is nonetheless vital that the Supreme Court judges clarify their approach on AI patenting. Given the UK government's increasingly AI-first attitude, the court must also be mindful of how to deal with inevitable future AI cases.
3 March 2023 by Amy Sandys
Stephen Thaler, creator of AI machine Dabus, has already lost three rounds in the UK. The latest blow came in September 2021, when the Court of Appeal upheld that an inventor can only be a ‘natural person’, and not a machine. Thaler’s argument is that Dabus itself should be named as the inventor and owner of EP 35 64 144 and EP 35 63 896, since it was the AI device which developed or ‘created’ the inventions beind the patents.
The Supreme Court must therefore clarify an issue which the UK patent courts have spent the best part of five years debating. Much publicity has surrounded Dabus, for example the EPO Boards of Appeal decision leading to experts demanding an amendment of the European Patent Convention. But Thaler also failed to get the EPO to grant the Dabus patents, which would have been a boon for his mission globally.
Since the UK is so far the only national court jurisdiction where the case has reached Supreme Court level, patent lawyers globally are looking to its judges to make a reasoned decision. However, on a national level, UK government policy is taking an increasingly AI-first approach. AI systems such as ChatGPT are also beginning to shape the daily life and work of its people. It is therefore imperative that the Supreme Court also considers how current case law could approach similar issues in the future.
For many observers, the Supreme Court’s judgment is a foregone conclusion. Especially given Dabus’ lack of success in the UK and globally, it seems unlikely that the highest UK court instance will now undo the two previous decisions, which are based on case law laid out in the Patents Act 1977. While the act stipulates that an inventor must be a named person, some Dabus supporters – as well as those with an eye on AI’s potential – suggest that an act written 46 years ago is not longer fit for purpose.
Similarly, the Court of Appeal hearing produced differing opinions among the judges. In judge Colin Birss’ view, Thaler has “complied with his legal obligations” in that he correctly declared who he believes to be the inventor. Although Birss agreed that an inventor must be a person, he noted that the case was essentially regarding “the correct way to process patent applications through the Patent Office.”
As such, Birss considered that Thaler should receive a patent for essentially following the procedures and processes correctly. This conviction is a key aspect of the case’s referral to the Supreme Court. Regardless of external perspectives, the case is opening up broader questions on the provision of UK patent jurisprudence.
Furthermore, when Marcus Smith, presiding High Court judge, first considered the UKIPO appeal, he also rejected Thaler’s appeal with a caveat. Despite Smith upholding the patent office’s view, he noted that Dabus itself was not incapable of developing a creative concept. Indeed, the judge went as far as to say that Dabus did ‘invent’ the technology behind the patents.
The case’s elevation to the Supreme Court is no doubt highlighting its importance in shaping the future of AI patenting. But lower-instance courts had also considered that Dabus might impact more than current talking points. In 2020, the Court of Appeal originally granted permission for Thaler to appeal the first-instance decision under CPR 52.6(1)(b), on the basis that an important principle is at stake.
In doing so, the UK judiciary has already acknowledged that this issue requires careful deliberation. Whether the patent community likes it or not, Thaler and Dabus have opened a conversation on how national courts should approach similar issues.
The general market view is that the Supreme Court is unlikely to overturn the UK’s previous two decisions. This will keep the country’s approach in line with global verdicts, with the USPTO, Germany, and the Australia and New Zealand patent offices, among others. South Africa remains the only country to have dissented from the prevailing view. In 2021, the South African Companies and Intellectual Property Commission (CIPC) granted Dabus a patent listing.
Perhaps most importantly, last year the EPO Boards of Appeal determined that AI cannot be named as the inventor behind a patent. It rejected Dabus creator Stephen Thaler’s contention that “the invention was conceived autonomously by Dabus” and that the appellant had derived the right to the European patent “by virtue of being the owner and creator of Dabus”. However, the decision did contain certain nuances which suggest future developments could make the EPO more open to granting patents for AI systems.
For example, the decision notes that “under Article 52(1) EPC any invention which is novel, industrially applicable and involves an inventive step is patentable.”
Similarly, in 2021 the German Federal Patent Court seemed to follow protocol in its decision that a listed inventor must be a natural person, even if the AI has identified both the problem and the solution. While it ruled that the listed inventor must be a natural person, however, its judges also determined that the AI system itself can be additionally named on a patent application.
Thus, it is not all black and white. The UK’s Court of Appeal is not the only legal body to have provided somewhat contradictory opinions. Neither is it the only court to indicate that a flexible approach towards Dabus is necessary when judges consider AI issues of the future. Yesterday’s hearing, therefore, should finalise the UK’s approach at its highest instance. But it also needs to consider how the judiciary might approach similar questions when they inevitably arise in the future, which now seems likely give the rise of software such as ChatGPT. For the latter, questions of ownership over text copyright are likely to arise.
For AI patent holders who are keen to understand the possibilites of patenting AI- and machine-driven inventions, a clear decision is necessary so that they can develop future strategies. On a judicial level, the Supreme Court must also show the country’s law makers that it is forward-thinking when considering ground less-well trodden. Perhaps this could ultimately lead to a revision of the Patents Act 1977.
After all, the UK government invested around £2.3 billion in AI initiatives between 2014 and 2021. It is also undertaking an AI strategy directed at business. Although following a consultation, the government’s current position is to maintain the law on AI as it stands, the outlook is unlikely to remain the same for long. Soon enough, a similar case will once again open the door to discussion and negotiation on the current scope for patenting AI inventions.