The European Union's "fragmented" regulation of artificial intelligence (AI) is hindering its own development?
Recently, Meta CEO Mark Zuckerberg, Ericsson President and CEO Börje Ekholm, and nearly 50 other company executives, researchers, and industry organizations have jointly written to the EU, stating: "Due to inconsistent regulatory decisions, Europe faces the risk of falling further behind in the age of artificial intelligence."
"This open letter is not intended to challenge the newly enacted EU Artificial Intelligence Act."
Partner at the British law firm Pinsent Masons, Nils Rauer, said: "The focus of concern is data.
Every AI-based application requires full access to data.
This begins with model training and continues throughout the product's entire lifecycle."
The EU's General Data Protection Regulation (GDPR), which has been in effect for more than six years, has become the "target of public criticism" in the joint letter.
The letter states that if hundreds of billions of euros are to be invested in creating generative artificial intelligence for European citizens, companies and institutions need a set of clear and consistently applied rules that allow them to use European data.
Associate Professor Zhang Longtian from the School of International Economics and Trade at Central University of Finance and Economics, in an interview with First Financial Daily, said that the EU's regulatory policies such as the GDPR provide important protection for data privacy and security, but to some extent, they also increase the compliance costs of enterprises and suppress innovation vitality.
He further explained: "Especially in emerging technology fields such as artificial intelligence, the free flow of data is a key factor in driving technological progress.
The heavy regulatory environment in the EU may lead to more restrictions on technology companies, thereby reducing competitiveness."
GDPR becomes the "target of public criticism" GDPR requires data processors to obtain the explicit consent of the data subject before data can be used and to anonymize it.
It has clear regulations on the mechanism for data cross-border transmission from member states to third countries, and the penalty for violations can be as high as 20 million euros or 4% of the company's global annual turnover, whichever is higher.
Tech giants such as Meta and Amazon have faced hundreds of millions of euros in fines for violating GDPR by transferring EU user data to the United States.
Research published by David S. Evans, a director at the consulting firm Berkeley Research Group, in May this year, showed that to meet the requirements of GDPR, EU companies have reduced their data storage by 26% and data processing by 15% compared to similar companies in the United States.
For companies, the average cost of data has increased by 20%.
The signatories stated that in recent years, the EU's regulatory decisions have become "fragmented" and "difficult to predict," and they said that due to the intervention of the EU's Data Protection Authority (DPA), there is great uncertainty about which data can be used for AI model training.
"This means that the next generation of open-source AI models and products and services built on it will not understand or reflect European knowledge, culture, or language," they said.
Previously, due to concerns about the risk of compliance with data usage, several tech giants, including Meta, which coordinated the letter, announced a delay in the entry of AI products into the European market.
In May this year, Meta announced plans to train AI using publicly available content from EU and UK users on Instagram and Facebook.
The non-profit organization European Center for Digital Rights (NOYB) immediately filed complaints with 11 EU member states, requesting the initiation of emergency procedures.
Subsequently, at the request of regulatory authorities in Ireland, the UK, Norway, and other countries, Meta announced the suspension of this plan and postponed the launch of its own large model Meta AI in Europe.
Recently, X also announced the suspension of training Grok using data from European users for the same reason and may face fines.
In addition, Apple announced in June that it would temporarily not launch three new AI technologies, including Apple Intelligence, in the EU market.
Google also postponed the launch of the AI chatbot Bard in the EU last year.
Zhang Longtian believes that the stance in the letter may have a limited impact on the formulation of EU policies in the short term.
He explained that the EU's policy-making process is complex and slow, involving coordination between multiple member states, interest groups, and legislative bodies.
Different interests and priorities make it difficult for the EU to quickly adjust the existing policy framework.
In addition, regulatory coordination within the EU often faces challenges, and there are differences in the implementation of specific policies among member states, which further delays the process of policy adjustment.
Currently, the development among EU member states in artificial intelligence is not balanced.
France and Germany have concentrated a group of the most promising AI startups in the EU, such as Aleph Alpha, DeepL, Mistral AI, Hugging Face, etc.
A report released by the European Court of Auditors (ECA) in May also showed that the two countries have the largest scale of investment in artificial intelligence, while four EU countries have not yet announced any AI strategy.
Turning to London?
In the open letter, the signatories also issued a warning: "Compared with other regions, Europe's competitiveness and innovation capabilities have declined."
According to the global listed company market value ranking website CompaniesMarketCap, among the top 50 internet companies with the highest market value in the world, China and the United States account for 39, and only Sweden's Spotify is on the list in Europe.
Nicolas Petit, a professor at the European University Institute in Florence, once said that since the birth of the Internet, Europe has no "superstar companies."
Digital Europe, a group representing the European technology industry, said in April that only 3% of the global AI unicorns come from EU member states.
Different from the EU's approach to strengthening regulation, the British government previously adopted a light-touch, principle-based, "support innovation" approach to AI regulation.
At the first Global AI Safety Summit held in November last year, then-British Prime Minister Sunak said: "The UK's answer is not to rush to regulate," and said "We will always encourage innovation, not kill innovation."
According to research by the venture capital firm Accel, London is already the largest generative AI center in Europe, with nearly 30% of emerging startups in Europe.
Companies such as Microsoft and Salesforce also announced this year that they will open new AI centers in London.
In this regard, Zhang Longtian said: "As the financial technology and innovation hub of Europe, London has strong academic and technical resources, which provide a good environment for corporate research and development.
At the same time, the investment of these companies will also attract more high-skilled talents into the European market, promoting knowledge sharing and technology transfer."
Meta has also recently announced the resumption of training Meta AI using publicly shared content from adult users on Facebook and Instagram in the UK to ensure that it reflects British culture, history, and idioms.
However, the new Labor government led by British Prime Minister Starmer has indicated that it will seek to enact appropriate legislation to make demands on those who are committed to developing "the most powerful AI models."
Due to the pressure to cut costs, a £1.3 billion AI investment plan formulated by the previous Conservative government has been canceled.
Cerys Wyn Davies, a partner at Pinsent Masons Law Firm, also said: "Companies need legal certainty and security.
These are the prerequisites for investing in the development of modern technology."
