EU parliament: Stop content theft by AI builders
With an overwhelming majority, the European Parliament supports a proposal protecting copyrighted works from AI poaching and predatory activities.
With 460 votes in favour, 71 against and 88 abstentions, the parliament adopted an own-initiative report that significantly raises the bar for AI companies using copyrighted material.
Simply put, they have to start proving that they have trained their models with explicit permission on intellectual property of the original creators.
There needs to be a system of licensing where owners of intellectual property can record who can do what with their work under what conditions. And who did. One condition could be that the AI factory pays to train its models on the creative work.
A condition could be that the AI factory pays to train its models on the creative work, coincidentally. A UK parliamentary committee published a comparable proposal last week with a binary choice of either weakening the laws around copyright, or strictly and clearly defining what can and cannot be done.
The report called on the European Commission to establish this licensing mechanism. The core of the proposal is clear: generative AI systems should be covered by European copyright, regardless of where the training took place.
This position could have major implications for tech giants that train their models outside the EU but then offer them in the European market.
The management of the new mechanism should be in the hands of the European Intellectual Property Office (EUIPO).
From organisations representing the technology sector, discontent can be heard. The Computer and Communications Industry Association (CCIA) immediately climbed into the pen and objected with: “The text and data mining (TDM) exception in the Copyright Directive allows developers to train their AI models on publicly available material unless rights holders opt out. They exercise that right.”
The parliamentary proposal “would impose a compliance charge on EU companies in many sectors, but risks hitting start-ups in particular. They will not be able to negotiate complex licences with big publishers and therefore be unable to enter the market.”
Members of lobbying club CCIA are predominantly US tech companies, notably Amazon, Google, Meta and X.
Emerce






Subscribers 0
Fans 0
Followers 0
Followers