Following a Call for Views earlier this year, the Government has now published its consultation on potential changes to the UK IP protection framework, seeking to ensure that the right balance is struck between encouraging the development and use of AI across the economy, whilst not undermining IP's wider role in promoting human creativity and innovation.
The consultation focuses on three specific areas:
- Copyright protection for computer-generated works without a human author
- Licensing or exceptions to copyright for text and data mining
- Patent protection for AI-devised inventions
The Government does not express any firm views as to its preferred approach in relation to any of these issues, as it is hoping to gather additional evidence during the consultation process, but reiterates that any measures put in place must:
- Encourage innovation in AI technology and promote its use for the public good
- Preserve the central role of IP in promoting human creativity and innovation
- Be based on the best available economic evidence
Copyright protection for computer-generated works
Under current UK law, and in contrast to the approach adopted in most other countries, copyright is available to protect computer-generated works (CGWs) where there is no human creator. The author of such a work is deemed to be the person by whom the necessary arrangements for the creation of the work were undertaken (this could be for example, the user of the AI tool), and protection lasts for 50 years from the date when the work was made.
The consultation seeks to understand whether this regime strikes the right balance in terms of incentivising and rewarding investment in AI creativity. For example, it notes that, whilst there have been some high profile instances of CGW artworks (such as, for example, Portrait of Edmond Belamy, creative by the collective Obvious), there has been no evidence in the UK of human creators facing significant competitive pressure from CGWs.
However, some have criticised the current law for being unclear and contradictory – a work, including a CGW, must be original to be protected by copyright, but the test for originality is defined by reference to human authors, and by reference to human traits such as whether it reflects their 'free and expressive choices' and whether it contains their 'stamp of personality'.
From an economic perspective, meanwhile, it has been argued that providing copyright protection for CGWs is excessive because the incentive argument does not apply to computers. The consultation notes, for example, that it appears from the position in both the US and EU (where CGWs are not protected), that there are sufficient incentives to create CGWs without needing copyright protection. Further, some argue from a philosophical viewpoint that copyright should be available to protect only human creations, and that granting protection for CGWs devalues the worth of human creativity.
In light of the above points, the consultation sets out three potential policy options:
- Retain the current scheme of protection for CGWs - retaining the status quo would be justified if the current approach is demonstrated to incentivise creation of new AI-generated works and investment in AI technology, without presenting unreasonable costs to users and human creators.
- Remove protection for CGWs – a drastic option would be to remove copyright protection for future CGWs, and to therefore limit copyright protection to human authors. This would be appropriate if it were considered that copyright protection for CGWs do not incentivise AI investment, or that it comes at unreasonable cost to human creators and users. This proposal would not impact on AI-assisted works (i.e., those made by a human with the assistance of AI), nor 'entrepreneurial' works such as sound recordings, films, broadcasts or published editions (as these works do not have an originality requirement).
- Introduce a new right of protection for CGWs, with a reduced scope and duration – the 'compromise' option is to create a lesser form of copyright protection for future CGWs (modelled on existing entrepreneurial rights such as sound recordings), which may be considered appropriate if more limited protection was found to present a better balance between rights holders and third parties. The Government doesn't make a firm proposal on the term of any such lesser right, but puts forward five years as an example.
Text and data mining
Text and data mining (TDM) is a crucial tool in the development and training of AI tools, and more broadly in many areas of research involving computational analysis. In 2014, a specific copyright exception was introduced in the UK relating to the making of copies for the purposes of TDM. Certain criteria must be met in order to rely upon the exception: in particular, it permits non-commercial research only (and does not apply to databases), and the researcher must have lawful access to the work (contract terms preventing the making of copies to carry out TDM are unenforceable).
Meanwhile, the EU's Digital Single Market Copyright Directive (which EU Member States were required to implement by 7 June 2021) has recently introduced mandatory exceptions relating to TDM. The first is focused on TDM for scientific research purposes. The second is more general and is applicable to all users of content to which they have lawful access, but it carries a more limited scope (including a rights holder opt-out).
The Government is considering five policy options in relation to TDM:
- Retain the current exception for TDM, unchanged: Under this option, updated guidance would be provided on the definition of non-commercial research and in relation to fair dealing uses.
- Improve licensing environment for the purposes of TDM: This option would involve implementing best practices in the licensing of rights for TDM through educational materials, model licences or codes of practice, in order to promote legal certainty.
- Extend the existing exception for TDM to cover commercial research and databases: This would allow a broader use of TDM in a range of research projects. Users would still require lawful access to the material.
- Adopt a TDM exception for any use, with a right holder opt-out: Under this option, any use of TDM would be permitted, and by anyone – including, for example, for business analytics, journalism and citizen engagement. This option is modelled on the EU exception noted above and provides rights holders with the right to opt-out individual works, sets of works or all of their works if they do not want them to be mined.
- Adopt a TDM exception for any use, with no right holder opt-out: This option is similar to Japan's exception for information analysis, and a new copyright exception shortly to be implemented in Singapore. Clearly, whilst this would be welcomed by researchers, and those training AI systems, it would not be popular with rights holders.
Earlier this year, the Court of Appeal rejected the argument, in its decision concerning the Dabus machine, that an AI machine can be an inventor under the Patents Act 1977. However, the Federal Court of Australia has recently allowed an application naming Dabus as an inventor to be granted, and it is possible that the UK Supreme Court will grant permission to appeal in the Dabus case (particularly as there was some disagreement in the Court of Appeal as to the correct approach to adopt).
Against that backdrop, the Government is considering four policy options in relation to patent inventorship and AI:
- Continue to require a human inventor: This would endorse the approach set out by the Court of Appeal: patents will only be available where there is a human inventor, and AI should not be named as the, or a, deviser of the invention. The consultation notes that this option will assist businesses operating internationally as it will provide consistency across important commercial markets.
- Expand definition of 'inventor' to include humans responsible for an AI system which devises inventions: Under this option, inventions devised by AI would be patentable, but the patent application would still need to name a human inventor - this would be the human who made the necessary arrangements for the AI to devise the invention. This could be those involved in programming, configuring, or operating the AI; or selecting input data or recognising applications of the output of the AI.
- Allow patent applications to identify AI as the inventor: As with the second option, inventions devised by AI would be patentable: but here, AI would be named as the inventor, or alternatively the requirement to name an inventor could be removed where the invention is devised by AI. The patent would be owned by the human closely responsible for the invention, not the AI itself. The consultation notes that this option might impact on gaining patent protection in those markets which only recognise human inventors.
- Protect AI-devised inventions through a new type of protection: Under this option, a new separate type of protection would be created, the first time such an approach has been adopted anywhere in the world. The new right would be similar to patent protection, but would grant more limited exclusive rights (e.g., a shorter term of protection), and could have different conditions for grant such as a stricter test for inventive step, or perhaps no test of obviousness at all. There is, of course, a risk that providing a more limited right for AI-generated inventions might lead to applicants not naming AI as the inventor, in order to obtain the benefit of the lengthier term of patent protection.
The consultation closes at 11.45pm on 7 January 2022. As with other recent UKIPO consultations, the analysis of the preferred policy outcome is going to be complex: there will be a range of potentially conflicting views from all stakeholders in the AI and IP ecosystems: rights holders, AI service providers, creators/inventors, and users, many of whom will also be taking into account the impact of the proposals on their global innovation and protection strategies.