According to the Financial Times,U.K.The government's culture secretary, Lucy Fraser, has announced that rules on transparency in artificial intelligence will be fast-tracked to help protect content creators.
She said the rules would mean AI companies must be more open and transparent about the data used to train modelsUsers can also choose which data can be used by AI companies for training, and creators whose works are used to train models will also be paid.
Image source: Pexels
Not only the UK, but the EU is also currently formulating similar rules under its AI Act. According to the Act,AI developers will need to provide a sufficiently detailed outline of the training content, and implement policies to ensure that their operations comply with EU copyright law.
As part of its response, OpenAI has reached agreements with data providers. According to previous reports by IT Home, OpenAI has recently announced cooperation with Reddit, Stack Overflow and the Financial Times to use each other's data to train its own models.
The report said that if the above rules eventually become law, it will encourage users who claim that their rights have been violated, but for users of AI models, it may also lead to a decline in the quality of AI output content.