UK Ruling Revitalises Discussions On Harmonizing AI And IP

The Dec. 20, 2023, U.K. Supreme Court decision in Stephen Thaler v. Comptroller-General of Patents, Designs and Trade Marks has reinvigorated ongoing discussions about how the developments in artificial intelligence fit within the existing intellectual property legislative landscape.

We will turn to the decision shortly, but it is helpful to first consider some of the U.K. policymaking background on AI and its relationship with IP.

The U.K. Intellectual Property Office Consultation on AI and IP Law

In October 2021, the U.K. Intellectual Property Office, or UKIPO, opened a consultation regarding three specific issues involving AI and IP under English law:

• Copyright protection for computer-generated works, or CGWs, without a human author;
• A licensing regime or exceptions to copyright for text and data mining, or TDM, which frequently plays a role in AI use and development; and
• Patent protection for inventions devised by AI systems.

By the end of June 2022, the UKIPO had received and considered the responses, and suggested potential changes to the law - although it subsequently changed its mind on one aspect, which we will return to.

Copyright Law for Computer-Generated Works

Under the law at the time, CGWs were recognized as being eligible for copyright protection. Copyright lasts for 50 years from the end of the calendar year in which the work was made, and authorship is given to the person or persons by whom the arrangements for the creation of the CGW were made.

The consultation explored whether copyright law should be amended to remove protection for CGWs without a human author. It decided to maintain the existing law, with possible future revisions based on emerging evidence.

Licensing or Exceptions for Text and Data Mining

The UKIPO considered introducing copyright exceptions or licensing for TDM.

In essence, TDM is the electronic scraping and analysis of existing material in quantities and speeds far in excess of human processing capabilities. It, therefore, inherently involves copying large amounts of material, which in itself may very often be protected by copyright.

The decision was to change the law and implement a broad copyright exception for TDM for any purpose, with safeguards for rights holders.

Patent Law for AI-Devised Inventions

The consultation evaluated whether patent law should be updated to allow patent protection for AI-generated inventions with no human inventor. The consensus was that an AI system is not yet advanced enough to be considered an inventor, but this position may change with future advancements in AI technology.

The Takeaways

In summary, the only change that the government decided to make at the time was to introduce an all-encompassing exception for TDM, significantly expanding the existing exception for noncommercial TDM. That decision was, though, short-lived — the government actually reversed its planned change early the following year over concerns that the rights of creators were not being sufficiently well safeguarded.

The government's response on Jan. 9 to the House of Commons' 2023 report, "Connected tech: AI and creative technology," confirmed this reversal and recommended that the government support the continuance of a strong copyright regime with licenses required to use copyrighted content in AI.

A further commitment has been made to develop a code of practice on copyright and AI, to enable the AI and creative sectors to grow in partnership. Further recommendations are expected with the response to the AI regulation white paper due later this year.

Copyright Comparison with the European Union

This volte-face on TDM leaves the U.K. with a regime that is friendlier to creators than the European Union's but less helpful to AI owners or deployers. In the EU, there is a general TDM exception to copyright and database right infringement, although there is an opt-out mechanism available to right holders.

Focus on Patent Issues

On the issue of patent law, it seems that there were some voices in the consultation, particularly from the pharmaceutical sector, that are keen to see the law move ahead of the technology and recognize the ability, or at least potential, of AI systems to create inventions. Concerns were voiced that, without a change in the law, inventions made in the future using AI technology might not be granted protection.

The majority view was against change, however, with the government acknowledging that "unilateral change … risks being counterproductive." Many respondents expressed the view that international harmonization would be crucial to the success of policymaking on this topic.

At present, the prevailing approach internationally is not to recognize AI machines as potential inventors, with the U.S., the EU Boards of Appeal and Australia being just three key markets aligning with the U.K. on this point.
 
To us, this thinking seems to be intuitively correct. There does not appear to be any overwhelming policy reason at present for amending the law to provide greater recognition of the contributions to the creation of IP made by AI systems.

One can imagine a situation in the future in which AI systems have a greater degree of autonomy and are perhaps even able to contribute financially to the IP ecosystem, for example, by paying for patent filing fees themselves and perhaps even by licensing their patents to third parties - or perhaps to other AI actors - with no, or at least minimal, human interference. In such circumstances, it would likely be sensible to review the law on the ability of such systems to create inventions.

We are, though, quite a long way from this reality, and current patent laws that enable the protection of inventions made with AI assistance would seem to meet current needs.

The Thaler Case

Fast forwarding to Dec. 20, 2023, and the Supreme Court's decision in Thaler v. Comptroller-General of Patents, Designs and Trade Marks, the court addressed the question of whether an AI system can be legally recognized as an inventor for the purposes of patent applications.

Stephen Thaler had filed applications for inventions created by an AI machine - DABUS - that he owned, listing DABUS as the inventor. The UKIPO had previously rejected these applications on the grounds that an inventor must be a natural person.

After the case had been heard in the High Court of Justice of England and Wales and the Court of Appeal of England and Wales, which both confirmed the UKIPO's finding, the Supreme Court ultimately upheld the decision, affirming that, under current U.K. law, an inventor must be an individual human being and not an AI system, and further that there was no right for Thaler to apply for and obtain a patent as the owner of DABUS.

The court's judgment hinged on the interpretation of the Patents Act 1977, which is structured around the concept of an inventor as a natural person. On this, the Supreme Court was clear: The identity of an inventor as a natural person goes to the heart of the system for granting a monopoly for an invention; it is not a tenable interpretation of the 1977 act that a machine can be an inventor. The court stated that "an inventor within the meaning of the 1977 Act must be a natural person, and DABUS is not a person at all, let alone a natural person."

The broader questions of whether technical advances generated by machines acting autonomously and powered by AI should be patentable or whether the meaning of the term "inventor" ought to be expanded to include machines powered by AI were not before the Supreme Court to consider - any change in this legal position would require legislative action.

As we have already seen from the 2021/22 government consultation, for now, the government has decided not to change the law on this. We expect continued international discussion and harmonization on this point, with a close eye being kept on technological development.

While, as noted above, we understand the current position on not conferring inventor status on machines, there seems to us to be a developing tension between the approach toward CGWs regarding copyright protection and AI-assisted inventing with respect to patents.
 
CGWs can be granted copyright protection whereby the author is the person by whom the arrangements were made. However, as confirmed by the Supreme Court in Thaler, patents must have a natural person cited as the inventor, and ownership or creation of the inventing machine — effectively the making of arrangements for the machine to operate — is not sufficient to confer a right to apply for and obtain a patent for an invention that the machine may devise.

It is interesting that Justice David Kitchin, who gave the leading judgment, noted that Thaler never argued that he was the inventor and used DABUS as a highly sophisticated tool. It was his view that, had he done so, the outcome of the proceedings might well have been different.

The "Connected tech: AI and creative technology" House of Commons Committee report, published on Aug. 30, 2023, indicates that the digital and tech sector contributed £150.9 billion ($190.9 billion) to the U.K. economy in 2019. The value and influence of this sector will only continue to grow, and its effective regulation will be critical.

Of course, this is not just a U.K. issue. The development of generative AI and intellectual property is being considered at a global level. For example, it was recently discussed by the World Intellectual Property Office at the eighth session of the WIPO Conversation.[6] This remains an area to watch with keen interest.

This first appeared in Law360, on 31 January