This month the EU introduced copyright laws that, for the first time, hold online platforms legally responsible for enforcing copyright. This move returns some level of online content ownership to producers such as musicians and film makers.
Unlike a hit record, however, you can’t claim ownership of the data that you generate everyday. Data is the highly prized raw material that allows engineers to ‘teach’ AI human skills such as discriminating between people and objects (critical for self driving cars) or finding a tumour in brain scan.
User profiling, targeted advertising and behavioural nudging: it’s hard to argue convincingly that any of these benefit the consumer or society, yet they have enormous commercial value. The individual is lost in the mix – no ownership, no voice.
advertisement 
Right now, AI-savvy companies can harvest all this data and turn it to profit with little or no regard for the wishes of those who produced it. This is a twin peril for the public – no benefits to us from our profitable data and no representation when it comes to deciding safeguards on how it can be used.
International gathering
In April, AI experts from around the world will gather in Dublin to advance technical standards for AI trustworthy. While technical solutions can help make AI more reliable and free from bias, they can’t alone ensure AI is used in our best interests. There are vast areas of vulnerability that commercial bodies show little interest in closing off and few levers that people and legislators can bring to bear.
GDPR offer some specific rights, but only for personal data, and not in the form of familiar ownership rights that can be traded or bequeathed to descendants.
Our rights over data, to the extent that they exist, are therefore grounded in a complex set of copyright and privacy legislation. The EU recognises this as an impediment to the free flow of data needed to develop a successful European data economy, but it also limits our ability to control our data.
One of the key obstacles to giving the data-producer (ie you) real rights is that we struggle to fit data into our traditional ideas of ownership.
Land can be inherited, royalties go to an individual or an estate. If your car is stolen you have documents to prove you owned it in the first place. However, with no meaningful definition of ownership for our everyday data, it’s a free-for-all for commercial interests to take without asking, and to use it in any way that suits them.
User benefits
User profiling, targeted advertising and behavioural nudging: it’s hard to argue convincingly that any of these benefit the consumer or society, yet they have enormous commercial value. The individual is lost in the mix – no ownership, no voice.
We need another approach, but one that can keep pace with the accelerating change of an artificially intelligent world in a way that centralised government regulation cannot.
However, we mustn’t throw out the AI baby with the data bathwater. Data-driven AI innovation can be a great economic and social benefit, provided it can proceed fairly.
Ownership definition
One proposal to is to create a new EU data ownership definition. How would this look? Very simply, a non-exclusive ownership right over any data for any party that can show evidence of having produced or contributed to that data. This is combined with an obligation on right-holders to share data on fair, reasonable and non-discriminatory terms.
Such a simple, universal data ownership right would enhance the flow of data between parties by clarifying rights and simplifying data sharing contracts. Critically though, it would also give right holders a legal lever to negotiate safeguards and benefits with the companies deriving value from their data.
However, the value of data comes only when companies combine it in large volumes from many sources. So the individual alone has very little power in negotiating with the large organisations that currently reap value most from our data. For example, if a dating app is providing you with a ‘free’ service it’s hard to force its provider to avoid any bias against a particular minority. Could data right-holders redress this imbalance of power by acting collectively?
Here again, EU law may be ready to play a role. As well as providing the basis for privacy right underpinning GDPR, the EU Charter of Fundamental Rights also addresses solidarity, including the right to collective bargaining and action.
I suggest that it is not such a leap of imagination for a new universal data ownership right to be combined with a right for data owners, the ‘workers’ in the new data economy, to bargain and act collectively.
Collective negotiation
Rights-holders of data feeding an AI system could collectively negotiate the conditions for data use and ensure transparent and accountable implementation.
An organisation collecting user data for AI development might even be compelled to negotiate with their elected representative. We all understand the role of union shop steward – such a role could conceivably be created for the data right-holders as a bloc.
This speculative proposal certainly requires much further analysis by legal scholars and practitioners. However, it may just offer the simple, scalable and fair path to control of how our data is used, perhaps to the rallying cry of no datafication without representation.
Professor Dave Lewis is associate director in the ADAPT Centre, Trinity College Dublin
Ltd
