The UK Intellectual Property Office (IPO) has published its response to its consultation “Artificial Intelligence and IP: copyright and patents”. The consultation considered three specific areas, namely:
- Copyright protection for computer-generated works (CGWs) without a human author.
- Licensing or exceptions to copyright for text and data mining (TDM), which is often significant in AI use and development.
- Patent protection for AI-devised inventions.
The protection of CGWs (i.e., where there is no human author of the work) is already recognised in section 9(3) of the Copyright, Designs and Patents Act 1988. The UK is something of an outlier in providing copyright protection to CGWs and its approach has given rise to some uncertainty, particularly around the concept of ‘authorship’.
The majority of respondents to the consultation favoured no change to the existing law and the IPO has followed suit. Their reasons for this are threefold: (1) there is no evidence that the existing position is harmful, (2) the use of AI is still in its early stages meaning that a proper evaluation is not possible, and (3) any change could have unintended consequences. However, the IPO will continue to monitor the position and remains open to new evidence on the effects of the existing regime.
Text and data mining
The UK’s existing TDM copyright exception is very narrow and it is unsurprising that it has come under scrutiny in view of the Government’s ambition under its National Data Strategy to be among the global AI superpowers (see our earlier post here). It applies only to research for a non-commercial purpose. Such a limited exception does not really support the TDM that is typically associated with the process, development and deployment of AI.
The IPO has therefore stated its intention to introduce a new TDM exception for both copyright and the database right, allowing TDM for any purpose (so including commercial TDM). Perhaps most significantly, right holders will no longer be entitled to charge licence fees for TDM or seek to ‘contract out’ of the new exception, which could have a significant impact on some existing data licensing business models. However, would-be data users will still be required to have lawful access to the text and data they wish to mine and so it remains open to right holders to select a platform on which to make their text and data available and charge for access (for example, a one-off fee or a subscription).
The IPO has elected not to make any change to existing UK law rather than, for example, expanding the definition of ‘inventor’, allowing an AI to be named as inventor or even introducing and new patent-like right to protect AI-devised inventions.
Many respondents to the consultation shared the view that AI is a tool used by human inventors and is not capable of inventing without significant human intervention. There was also a view that this issue is best addressed/harmonised at an international level, with the IPO identifying the actions it proposes to take in that regard.