ALPSP has published its response to the IPO's recent announcement:

Dear Mr. Adam Williams, Director of Business & International Policy at the IPO,

We are writing on behalf of ALPSP members and in response to the IPO’s recent announcement regarding the introduction of an extended text and data mining (TDM) copyright exception. We have significant concerns about the impact this exception may have across the whole creative industry landscape, but in particular how this will affect the smaller publishers represented by ALPSP.

We urge the IPO to reconsider its approach as soon as possible. The threat of an unrestricted exception is creating confusion for rights holders, tech companies, researchers, investors, and authors as they reconsider what they should invest in how protected works may be accessed and utilized for text mining purposes. Our members are constantly innovating on how they may best disseminate their content, whether that be through Open Access strategies or new types of licensing agreements. This has created a healthy AI industry in the UK where content providers work in partnership with AI/TDM organizations ensuring advances are based on verifiable and accurate information that is fit for purpose. We believe cutting publishers out from the AI industry’s future would be detrimental to this industry’s growth and to the government’s goal of making the UK a global technological leader.

Publishers of all sizes have developed a host of successful licensing strategies and policies across the range of TDM/AI/machine learning activities for both commercial and non-commercial TDM that are working well. TDM licences may exist in isolation, or in a bundle of products including curated data transfers and feeds. This helps to ensure TDM activities are based on the scientific record, or Version of Record (VOR) of an article, which in turn safeguards a range of interests including adherence to ethical guidelines. Licences also ensure that the content TDM relies upon is distributed, stored, and reutilized in a manner that does not prejudice either the author or the rights holders legitimate economic or moral interests; we have serious concerns about what type of industry would develop without those partnership agreements.

It is our understanding that commitments have already been made to re-examine the scope of this exception and for the IPO to engage in a full market assessment of the impact this exception could have on the creative industries. We support this proposal as it will show that publishers are already working strategically and comprehensively with the technology sector in this area. AI is a novel technology and as such there is not a plethora of data available, however we have a host of explanatory data on the ways in which publishers’ partner with the technology sector that we believe would help demonstrate the significant impact such an unrestricted exception could have on our sector.

Our members are also concerned as to the proposed scope of this exception. From our understanding, the IPO is proposing a novel exception where the only requirement (and therefore the only safeguard against abuse) for a TDM activity to occur would be “lawful access.”

• May lawful access be contractual, or does it only refer to technological efforts to limit accessibility?
• Is lawful access personal, or may organizational affiliation be enough?
• Does lawful access need to be perpetual, or may it be temporary?
• May copyright exceptions or limitations be relied upon to confer lawful access.

These are just a few preliminary questions that are causing great debate amongst our members, many of whom do not have IT departments to facilitate access and/or handle problems caused by server overload if the term ‘legal access’ is broad and general.

Our members of all sizes are engaged in licensing their content for TDM and have developed strong licensing practices. These licences clearly define how the TDM activity may occur, and our members have a range of different licences for different types of activity. They report their approach may be influenced by:

• Amount of data used in TDM/how data is stored and for how long?
• Amount of other copyright protected work, including databases, literary; artistic; computer generated works; audio visual works which may be reproduced as part of the TDM activity. This potentially involves many copyright interests across global markets.
• How TDM may occur: often publishers will feed TDM developers’ content directly to ensure smooth functioning of as a service as part of the licence.
• What output the TDM may include. Different TDM developers have different needs when it comes to showing the results of their mining activity. This may include wishing to reuse text, graphical, or audio extracts; and
• Dealing with any inclusion of potentially sensitive data or information.

Our members are troubled that this consultation ignores the resources and investment provided by rights holders to enable TDM to grow in a sustainable partnership with content creators.

Our members are also concerned that this exception would undermine the strong advances being made in Open Access (OA)/Open Science. STM publishers are rigorously engaged in expanding OA publications, and such publications have been through a process whereby all stakeholders are informed and aware of the implications of publishing OA. If tech companies do not wish to partner with publishers to license works for TDM activities, we believe they should be encouraged to look at OA material.

We are also worried about several of the conclusions reached in the consultation. For instance: ‘although factual data, trends and concepts are not protected by copyright, they are often embedded in copyright works. Data mining systems copy works to extract and analyze the data they contain.’ We believe this does not represent the full extent of TDM activities which often analyze not just the data but the expression of that data, whether that be in pictorial or linguistic form. The protection of this expression is at the heart of copyright law, and, for example, particularly around sensitive data that may be contained in photographs, there is real concern over how these works may be treated. Rights holders are best situated to regulate how their works may be reused and impose necessary safeguards which this exception threatens.

The consultation also recognizes other territories with TDM exceptions (EU, Japan, Singapore, US). All of these exceptions are restrained either by rights holder opt outs, the Berne three-step test, or the analysis of fair use factors. The UK’s proposal would be novel in having no limitations in copyright, and we fear this would chill rather than promote innovation.

We also have fundamental questions as to whether this proposed exception satisfies the Berne three-step test, or whether this would put our copyright regime in jeopardy as being out of line with global norms. We ask that a thorough assessment of the effect this legislation would have on market sectors be conducted as a matter of priority.

Our members report that the TDM landscape is complex. TDM developers have a range of interests, financial backing, and products that they put onto the market. The way these products integrate with protected works is diverse, but this is a growing and innovative market that publishers are actively engaged in. There does not appear to be any market failure, but it is this partnership that is allowing the UK to be a continued source of safe, reliable, and trusted innovation. The threat of this exception has put real brakes on growth, and we urge the IPO to announce a review as soon as possible.

An unrestricted exception is deleterious to innovation. It is hard to envisage how this exception would be technologically neutral, or how it would apply to changes and developments in this field. It does not promote rights holder engagement or investment in ensuring AI/machine learning activities are conducted in a rigorous or scientific manner. This discouragement for key stakeholder participation in the development of this field is particularly concerning considering how nascent AI/TDM technology is. We do not believe the current TDM regime needs to be changed. We do not see any evidence of market failure. Quite the contrary; we see a market which is functioning well and developing with the market.

We believe innovation has thrived in the UK not in spite of our strong copyright regimes, but because of them. Currently, publishers and technology companies work in alliance to ensure TDM activities grow in a sustainable fashion, and we appreciate your work in facilitating the furtherance of this field. We think you would find it valuable to learn about the practical ways our members of all sizes are dealing with AI/TDM/machine learning and welcome any opportunity to discuss such with you.

If you would like to discuss any aspect of this letter in more detail, then please do let us know. We would be happy to assist with next steps.

With kind regards,

Wayne Sime

Chief Executive
The Association of Learned & Professional Society Publishers (ALPSP)

Categories: data, licensing, rights