Europe flies to introduce AI regulatory framework with copyright protection at its core

On Thursday, members of the European Parliament reached a tentative political agreement on a proposal for The AI Act, which is expected to be voted on in key committees on May 11, with a plenary vote in mid-June. If passed, the bill would be the world's first comprehensive AI law.

However, the draft bill has not yet been finalized and may take several years to take effect.

It's worth noting that the speed of this agreement is "rare" because the European Parliament has been criticized for its slow decision-making.

According to Reuters, the small group of members quickly finalized what could become a landmark piece of legislation in just 11 days.

Europe's action is leading the way, and in its wake, other countries are beginning to act.

Today, the G-7's digital and technology ministers discussed the regulation of artificial intelligence in Takasaki, Honshu Island, Japan, Kyodo News reported.

Japan's Digital Minister Taro Kono said he hoped all parties could reach a consensus around the topics discussed at the meeting. the G7 ministers hope to adopt an action plan to promote the responsible use of AI, calling for stakeholder participation in exploring international standards.

Copyright protection

The bill reportedly centers on the copyright of materials used by AI companies to train their models.

Controversy over copyright infringement has surrounded AI companies in recent months: image publisher Getty Images sued Stable Diffusion for using copyrighted photos to train its system. openAI has also faced criticism for refusing to share details of the datasets used to train its software.

On April 17, more than a dozen MEPs involved in drafting the legislation signed an open letter and urged world leaders to hold a summit to find ways to control the development of advanced artificial intelligence.

That same day, two of the dozen MEPs proposed a change: companies with generative AI systems must disclose any copyrighted material - books, photos, videos, etc. - used to train their models, Reuters said, citing sources.

Another conservative MEP proposed that the companies must seek permission from rights holders before using the data, a proposal that was deemed "too restrictive and could hinder the development of the emerging industry.

In the following week, the European Parliament tabled proposed laws after discussing the details, with one MEP telling Reuters that they could "force the industry to reach an uncomfortable level of transparency:

There have been calls both inside and outside the Parliament to ban ChatGPT or classify it as high risk. The final compromise is in favor of innovation, as it does not classify these models as "high-risk," but imposes requirements for transparency and quality.

Significantly, the proposal has cross-party support. According to one lawmaker who told Reuters:

I must admit that I was really surprised that we agreed so easily on the content of these texts.

It shows that at this point in time there is a strong consensus and a common understanding of how to regulate.