ChatGPT, the large language model created by OpenAI and deployed in 2020, has faced criticism from regulators, academics and industry professionals across Europe. The latest to add their voices are a coalition of German trade unions representing 140,000 authors and performers, who have called on the EU to impose strict guidelines for the AI’s use of copyrighted material.
Verdi and DGB are two of the 42 trade organisations that signed the open letter, along with other associations for photographers, designers, journalists and illustrators. Referring to the problems arising from the material used to train the model, the signatories highlighted the privacy and accountability issues, as well as the need for fair remuneration for copyright holders.
Another landmark copyright challenge came in January, when Getty Images filed a suit against Stability AI for using over 12 million of its photos and captions without permission. The seemingly never-ending issues over copyright infringement have pushed Sam Altman, the CEO of OpenAI, to announce that the company is looking for alternative strategies to its language model.
GPT-4 was the latest model trained by OpenAI and cost around $100 million dollars. Its size accumulated over a trillion words, hugely amplifying the development of machine learning technology. Experiencing diminishing returns on scaling up models, OpenAI is looking for more efficient algorithms.
The upcoming EU Artificial Intelligence Act is likely to set a global regulatory standard, but it will have to be updated continually to address issues like copyright infringement. It is up to the legislators to find a balance that both keeps intellectual property protected and stimulates innovation in Artificial Intelligence.
