- BREIN, a Dutch copyright enforcement organisation, removed a linguistic data set used to train AI models.
- Danish Rights Alliance removed copyrighted material used for AI model training without permission.
OUR TAKE
Just as everything in society should be regulated, the training of AI models requires formal, legal, specialised and authorised content from a permitted channel. BREIN’ s move highlights the importance of respecting copyright laws in AI development and the need for transparency in AI model training.
–Lily,Yang, BTW reporter
What happened
A Dutch copyright protection organisation, BREIN, has removed a massive language dataset used for training AI models after it was found to include unauthorised content from books, news websites, and movie and TV show subtitles.
The dataset, which comprised Dutch-language information, was gathered without permission. Bastiaan van Ramshorst, the director of BREIN, stated that it is unclear whether the dataset has been used by AI companies or to what extent.
He noted that the European Union‘s AI Act will require AI companies to disclose the data sets used for training their models. For instance, Danish Rights Alliance asked a large dataset called “Books3” to take down last year. Similarly, in the US, the legal challenges of using copyright material to train AI models faced by AI companies, such as Microsoft.
Also read: Explained: How Telecom Regulator Is Cracking Down On Spam, Fraud Calls
Also read: Telecom fraudsters target younger generation
Why it’s important
The action of BREIN and European Union’s AI Act, as well as the challenge of Microsoft highlights the ongoing debate surrounding copyright infringement in the development of AI models.
The removal of the large language dataset by BREIN underscores the importance of respecting intellectual property rights. It is crucial for AI companies to ensure that they have proper authorisation to use data from copyrighted sources to train their models.
The impending European Union’s AI Act and the legal challenges faced by AI companies in the US emphasise the need for transparency and accountability in AI development. As AI continues to advance, it is essential to strike a balance between innovation and copyright protection.
