[On December 20, the UK Supreme Court affirmed its previous decision to deny registration to inventions by Dr. Stephen Thaler’s AI DABUS, holding that an AI software cannot be listed as an inventor. SpicyIP intern Vedika discusses this development. Vedika is a third-year B.A.LL.B. (Hons.) student at National Law University, Delhi. Her previous post can be accessed here.]
On December 20, 2023, the UK Supreme Court ruled against granting patent protection to two inventions whose investor was listed as DABUS (Device for the Autonomous Bootstrapping of Unified Sentience). Stephen Thaler, a US-based computer scientist had filed for patent protection of the AI-generated inventions in multiple jurisdictions, including India, in 2019.
The UK Court affirmed its previous decision to deny registration of the AI software as the inventor, stating that an inventor can either be a human or a company, but not a machine. The application stated that the invention was made by DABUS acting autonomously, and that Thaler was entitled to file for the patent by virtue of being the owner of the “creativity machine.” The Court reiterated that no human inventorship had been established, which was a requirement under the 1977 Act. Interestingly, this comes after another recent judgement where a London court ruled that an artificial neural network is patentable.
Thaler also received a similar rejection of registration from the US Patents and Trademarks Office earlier this year, where a the same ground of lack of a human inventor was relied on. As discussed here, South Africa and Australia are the only jurisdictions where DABUS has been registered as an inventor. While South African patent law does not define ‘investor,’ the Australian Federal Court held that an AI software may be considered an agent, which can be an inventing person or thing.
Thaler had also applied for registration of the patent in India (Application Number 202017019068, filed 05.05.2020). As seen from the screenshot below the application is still pending, though objections have been filed against registering an AI software as an inventor.
Unlike many other patent offices, the Indian Patent Office (IPO) has not issued AI-specific guidelines, and AI-related inventions are still assessed on the basis of the Computer-Related Inventions Guidelines of 2017. As in the case of copyright, patenting AI-generated content poses two main questions: can an AI be a sole or a joint inventor? If not, can the human who owns or controls the AI software be rightfully called the inventor(/creator), when the AI tool does, in fact, autonomously create the invention? The government is indeed looking to address these concerns through policy and legislative exercises discussed here. However, from the documents out so far, it is not clear how much the government has thought through these situations with particular focus on IP issues. While the use of AI itself needs much legislative attention, it is imperative for the government to clearly identify the interface between AI and IP and proactively work to address it. Arguably, AI has been around long enough for its repercussions to be felt across different industries and areas of law, and it might be apposite for the IPO to now lay down guidelines in light of the increasing reliance on AI for innovation in various industries (as recommended by the Parliamentary Standing Committee here (pdf)). Nevertheless, it may be argued that a much more comprehensive analysis of the impact of AI is required before jumping to regulate it. Let us know your position in the comments below.