The use of artificial intelligence programs has the potential to facilitate medical diagnoses, drug development, boost our knowledge in physics, astronomy… But it can also create major conflicts, so many as to lead different countries to propose a pause in its development or prohibit its use, as happened in Germany or Italy recently. And one of the biggest concerns is linked to copyright.
The European Union’s head of technology regulation, Margrethe Vestager, has declared today that the bloc she leads will probably reach a political deal this year that will pave the way for the world’s first major artificial intelligence (AI) legislation.
In an interview with the Reuters agency conducted during a meeting of digital ministers of the Group of Seven (Canada, France, Germany, Italy, Japan, the United Kingdom) in Takasaki, Japan, Vestager suggested legislative measures for the use of tools of AI, such as “labeling obligations for AI-generated images. There was no reason to hesitate and wait for legislation to pass to speed up the discussions needed to deliver the changes across systems where AI will have enormous influence.”
The measures are just the conclusion of a preliminary agreement that had already been reached on the EU’s Artificial Intelligence Law. From the moment the law enters into force, companies that implement generative artificial intelligence tools, such as ChatGPT, must disclose any copyrighted material by copyright used to develop their systems.
This provision was a late addition drafted in the last two weeks, according to a source familiar with the discussions. Some committee members initially proposed prohibit the use of copyrighted material to train generative AI models, the source said, but this was dropped in favor of a transparency requirement.
Contrary to what happened with other technologies, such as genetic engineering, the internet itself or nuclear energy, the European Commission began to drafting the AI Act almost two years agowhen it was an emerging technology (at the time of the launch of the OpenAI chatbot ChatGPT) and it was not, like today, on everyone’s lips almost constantly.
Members of the European Parliament agreed to take the draft to the next stage, the tripartite dialogue, during which EU lawmakers and member states will discuss the final details of the bill. At the moment and according to the proposals, the AI tools will be classified according to their perceived level of risk: from minimal to limited, high and unacceptable. Areas of concern could include biometric surveillance, the spread of misinformation, or discriminatory language.
“Contrary to conservative wishes for increased surveillance and leftist fantasies of excessive regulation,” explained Svenja Hahn, a German Member of the European Parliament, “parliament found a strong commitment that would regulate AI proportionately, it would protect the rights of citizens, encourage innovation and boost the economy.”
Fred Havemeyer, an analyst at the stock firm Macquarie and a specialist in technology, agrees with Hahn’s statement: “The EU proposal is one that has been made tactfully, rather than a ban-first, ask-questions-later approach proposed by some ”. Although the tools of high risk will not be prohibited, says the new law, those who use them must be very transparent in their operations. It will be necessary to see if the contravention of the norms only generates fines or produces consequences since this topic has not yet been addressed.