2024-09-14 11:00:08
Its report on this point, which has provoked opposition from Paris and companies such as Mistral or Meta, is expected at the end of 2024, before the world AI conference in Paris, in February 2025. Ms. Bensamoun is also authorized to give consultation on Business models between AI producers and content owners (press, publish, music, etc.), another delicate issue, while the first payment agreements are set between OpenAI and the Financial Times, News Agency (Wall Street Journal…), Medium Speed (The countryor, in France, Global.
Pwhy is AI shaking copyright?
Copyright was born from a technological change: printing. And all the revolutions of technology have shaken: the Internet, the mobile web, streaming… Today, it is the turn of AI to oppose it. Indeed, the original AI, which produces words, images or sounds, needs to be trained on quality data. And culture and media content is quality data. Therefore copyright is part of the value chain of AI design and must be valued.
How can we reconcile the two conflicting principles of the AI Law: the obligation for AI manufacturers to provide a detailed summary of training data and their right to invoke trade secrets to protect their knowledge?
The important thing is to get out of advertising and realize that there is a value chain. We pay for all the inputs: computers, talent, electricity, etc. The content used to train the AI also has a price. The problem is that the AI Act requires that this data collection be available to the public. It may be interesting to put it in the middle [par exemple, le Bureau européen de l’IA] to keep, if necessary, confidential.
That being said, we cannot invoke trade secrets to exempt ourselves from the application of the law. Also, this concept is mentioned in the explanatory notes [des annexes] of the text, but not in the provision itself, having legal force.
#Content #train #price