Data intermediaries
One of the key objectives of Prime Minister Keir Starmer is to “strengthen the role of trade unions in our society”.[1]
There is robust evidence that a strong trade union presence in the workplace leads to huge benefits, including higher pay for workers, less inequality, improved responses to industrial change and better productivity.[2]
This is why the government’s Plan to Make Work Pay is a central part of its mission to overhaul the economy after a long period of lacklustre growth and poor wage rises.
The Employment Rights Bill being considered by parliament introduces a set of measures that improve workers’ individual rights. These provide a crucial underpinning. But it also strengthens their collective voice on the understanding that workers on their own have little power in the workplace and are stronger when working together.
This analysis should be central to the government’s approach to issues about the use of data.
The workplace is evolving rapidly with new technology affecting the jobs that people do as well as the way that people are recruited, work is allocated to them and their performance is monitored.
A 2016 survey of nearly 7,500 workers found that while 87 per cent agreed with the statement “I am keen to embrace technology and maximise its benefits”, and 73 per cent agreed that technology would improve productivity, fewer than one in four (24 per cent) said that their employer gave them a say in how technology affects their work.[3]
In this environment, it is crucial that individual workers have rights to data, protection from employer misuse of data and are informed about these rights.
But the same analysis that acknowledges disparities of information and power in the workplace in employment rights should be applied to data rights.
For example, algorithms are often used to allocate work and determine payment for work. But individual workers often struggle to access the data compiled on them, and the algorithms used to process it via existing limited individual rights such as Subject Access Requests. And even if they can obtain this information, they have little ability to interpret that data in isolation while the employer is able to analyse collective data at scale. A far more effective method would be for workers and their unions to have the right to request that information so it can be compiled and analysed in collective data trusts.
There is a particular challenge in the platform economy. Often operators exploit clauses allowing workers to allow substitute workers to work on their account. This can allow the employer to avoid treating the workers as an employee or limb-b worker. It can also lead to the workers being denied data protection rights, such as the right to make a personal data access request. Therefore, they can be denied their rights.
The government is already committed to reforms to employment status. The TUC urges the government to move swiftly to better secure the boundary between worker status and self-employed status so that all workers who require the protection of the law, including access to data rights, will receive it.
In addition to such individual rights, workers should have a positive right to “data reciprocity”, to collect and combine workplace data so as to better understand the ways in which new technologies are being and can be used in the workplace, and to take advantage of this information for members and workers who consent. It is largely workers who generate the data, so data shouldn’t solely be used to the advantage of the employer.
As a starting point, we strongly urge ministers and civil servants to consider the measures set out in Section 8 of the model Artificial Intelligence (Employment and Regulation) Bill published by the TUC last year[4].
This includes measures that would give a trade union a right to be provided with the data collected by an employer that relates to its members that is used or is proposed to be used by the employer for artificial intelligence decision-making.
We also urge the government to consider ways of bringing discussion of work organisation and new technologies within the scope of collective bargaining. Research shows that unionised workplaces are more likely to employ ‘high performance’ methods of work organisation, such as team-working and problem-solving groups, than non-union workplaces.[5]
[1] Starmer, K. (2022). Speech to TUC Congress 2022. www.tuc.org.uk/speeches/keir-starmer-tuc-congress-2022
[2] TUC (2019). A Stronger Voice for Workers, TUC www.tuc.org.uk/research-analysis/reports/stronger-voice-workers
[3] Smith Institute (March 2016). The productivity puzzle: a view from employees
[4] TUC (2024). Artificial Intelligence (Regulation and Employment Rights) Bill www.tuc.org.uk/research-analysis/reports/artificial-intelligence-regula…
[5] Askenazy, P. and Forth, J. (2016). “Work organisation and HRM – does context matter?”, in Amossé, T., Bryson A., Forth, J. and Petit, H. (eds) Comparative Workplace Employment Relations: An Analysis of Britain and France, Palgrave Macmillan
Lack of awareness and understanding
We understand from our affiliate trade unions that there is a low level of awareness of all data subject rights under the UK General Data Protection Regulation. Not only this, but even where workers are aware of the rights available to them, they do not understand how these rights operate due to a widespread lack of guidance and education about data rights. For example, they are unlikely to understand what type of data is portable, or how they would arrange this to take place.
Whilst trade union officials have experience of assisting workers with data subject access requests, as well as working alongside employers to carry out data protection impact assessments. 1 We have not had any examples reported to us of workers exercising data portability rights.
In addition, there is very low awareness amongst workers of what data is being collected about them at work, and how this is being used. Prospect union surveyed 7,500 workers in 2020 and found that 48 per cent of workers are not confident that they know what data their employer collects about them, and 34 per cent are not confident that their employer will use their data in an appropriate way.2
There is a significant need for more guidance, education and awareness raising about the nature and value of data at work, how this is being used, subject rights under the UK GDPR and how these relate to the workplace.
Data rights also have renewed relevance in relation to the use of worker data for training AI models, or repurposing human creative work.3 This makes the need for awareness raising, education and guidance all the more urgent.
Not-for-profit data sharing
We advocate strongly for the importance of the support and development of data intermediary models based on a not-for-profit model, and for trade unions to be supported in their potential to act as data intermediaries.
The Drivers Seat app (www.driversseat.co), now part of the Workers Algorithmic Observatory ( https://wao.cs.princeton.edu/#about) is an example of a cooperative based project through which drivers and delivery workers shared pay and working time data in order to achieve better terms and conditions.
Worker Info Exchange is an organisation that demonstrates the significant potential for worker-focused organisations to act as data intermediaries: WIE makes data subject and portability requests on behalf of workers, and then collectivises this data for the purposes of advancing worker interests.
There are also examples of local authority bodies such as taxi licensing commissions attempting to collectivise driver data in order to establish better rates of pay and working conditions, For example, New York City Taxi Commission has attempted to obtain worker data from Uber drawing on a customer data sharing law which has been subject to litigation.
Commercial intermediaries
The TUC is concerned that any provisions relating to data portability in a contract of employment may be distorted by the power imbalance inherent in the employment relationship. For example, if consent is sought from an employee in the context of the employment contract for their data to be transferred to a commercial data intermediary, we query whether truly authentic consent can be obtained.
Therefore, where employers propose workers to consent to data portability to third parties, we advocate for the use of data intermediaries in the employment relationship to be subject to collective bargaining, so that the appropriate frameworks can be put in place in a collective agreement. This will also ensure that fundamental standards for the treatment of worker data by intermediaries are clearly set out in union negotiated agreements.
Collective data rights
The TUC is concerned that the Data Use and Access Bill is a missed opportunity to establish new digital rights. The Bill could, for example, have included provisions facilitating trade unions with enhanced collective data rights, and the ability to collectivise worker data as set out in the TUC’s model AI Bill (see below).
Legislation has not kept up with the emergence of data as a commercial product, with heightened value as the basis for training of AI models. Whilst technology companies and others are able to profit from the extraction of collective data, including from the employment context, individuals do not have reciprocal rights to benefit from their data, and the legal rights that are offered tend to be focused on individual rights, rather than collective rights involving institutions such as trade unions.
In a workplace context, this mismatch between reality and rights is magnified due to the power imbalance in the employment relationship, especially where trade unions are not recognised. For example, in Amazon fulfilment and distribution centres worker data is used to algorithmically calculate stringent targets for the workforce. However, workers are not able to know the targets themselves, nor the means they have been calculated. Workers are then subject to performance management and termination based on targets they do not have access to. Similar processes occur in the ‘gig economy’ where worker generated data is central to the opaque and highly changeable calculation of pay and conditions of work.
The imbalance of power over data – personal and non-personal - results in employers and technology companies benefitting from worker data in multiple ways, including as AI training data, without there necessarily being reciprocal benefit. For example, TUC affiliate Equity has a high-profile campaign illustrating the routine use of creative workers’ data and content to train AI models, without notice, consent, accreditation or remuneration. 4
The TUC has also recently published a manifesto for AI and creative workers, setting out the risks of unauthorised use of creative work for training data and the measures needed to guard against this, and to secure fair remuneration, including for historic abuses of copyright: https://www.tuc.org.uk/research-analysis/reports/artificial-intelligence-creative-workers.
However, we stress, unauthorised profiteering from worker data is not restricted to creative work, but also extends to a wide range of information about workers, their activities and behaviours.
We highlight the vulnerable position of workers and the self-employed who do not benefit from a contract of employment, yet whose data is harvested by platform companies, and others who are benefitting from their services.
We believe that regulation should be based on the understanding that data flows must be reciprocal. This means that employees and workers should be able to access all data flowing from them, collectivise it and use it to their own benefit. This can only work if workers also have full transparency over how their data is being used, and we believe, if they have the support of trade union assistance to redress the imbalance of power over data.
In our view, the ability to collectivise worker data could assist with identifying inequalities at work, including pay inequalities, as well as allowing the exposure of discriminatory, unethical and unlawful AI-powered decision making.
Trade unions are best placed to facilitate data aggregation and analysis from the perspective of workers, as well as potentially offer access to data scientists and applications to analyse this data.
We reproduce below the proposal we make in Part 8 of our model AI Bill for fair data use, in accordance with which trade unions would have the right to be provided with all member data an employer is using for AI decision making. 5
Fair data use
In accordance with this section, a trade union has the right to be provided with all the data collected by an employer that relates to its members that is used or is proposed to be used by the employer for artificial intelligence decision-making, to the extent that it is reasonably practicable.
Where a trade union makes a request in accordance with this section the employer shall provide the data in an accessible form within 2 months of the date of the request.
Except where the trade union’s members expressly agree in writing the data shall be provided in an anonymised form.
The right in subsection (1) shall not apply in respect of any member of the trade union who notifies the employer in writing of their objection to its collection.
The trade union shall notify the employer of the type of data it seeks and the date range to which the request relates; the range of data shall commence no earlier than 52 weeks prior to the request.
The right conferred by subsection (1) shall be enforceable by complaint to the employment tribunal in accordance with subsections 7 to 10 below.
A complaint brought under subsection (6) shall be commenced within 6 months of the last date by which the employer should have complied with the request.
It shall be for the employer to prove that it was not reasonably practicable to provide any particular data.
If the employment tribunal finds that the complaint is to any extent well founded it shall make a recommendation in accordance with Part 11.
The Secretary of State shall make regulations by order as to the manner in which a request under this section is to be made, the form by which an employee may lodge an objection to data being shared, what is reasonably practical, and the steps to be taken to preserve the anonymity of the data.
Support for data intermediary status
There is an absence of guidance relating to the specific challenges for not-for profit organising in becoming data intermediaries. For example, how to implement the legal structures and systems that may be needed in order to hold and process the data in a complaint manner, with minimum exposure to liability.
This type of structure change requires financial and human resource, as well as expertise, that may not be available to civil society and not for profit organisations. Lack of funding is a key barrier to trade unions adopting more active programmes of work to develop a role as data intermediaries.
- 1 For example, see this guidance from Prospect union: https://prospect.org.uk/about/data-protection-impact-assessments-a-union-guide
- 2 Prospect: Future of work, technology and data. https://prospect.org.uk/future-of-work-technology-and-data/
- 3 For example, the issues of the right of erasure and data access rights, are outlined in a report called AI: know your rights by Equity union: https://www.equity.org.uk/advice-and-support/know-your-rights/ai-toolkit/ai-know-your-rights.
- 4 http://www.equity.org.uk/campaigns-policy/stop-ai-stealing-the-show
- 5 TUC (2024). Artificial Intelligence (Regulation and Employment Rights) Bill http://www.tuc.org.uk/research-analysis/reports/artificial-intelligence-regulation-and-employment-rights-bill
Stay Updated
Want to hear about our latest news and blogs?
Sign up now to get it straight to your inbox