Experts say that China’s AI development shows that regulation does not have to be a barrier for innovation.
DeepSeek, a Chinese artificial intelligence large language model, has taken the US by surprise by quickly challenging dominant players such as OpenAI’s ChatGPT, Google’s Gemini and Meta’s LLama. Experts believe that this can also be a wake-up call for Europe’s AI innovators.
DeepSeek, a China-based company, was founded in July 2023 by Liang Wenfeng, an information and electronics engineer, and is the first of such advanced AI system which is available to users for free.
It comes after the rise of OpenAI’s ChatGPT which was first launched in 2022, and quickly became synonymous with the development of genAI.
In Europe, French start-up Mistal AI is the only real competitor to the US and Chinese giants. In 2024, it announced a new large language model set to rival OpenAI’s ChatGPT. However, the company previously complained that Europe lacks the data centres to train artificial intelligence models that match the current demand.
According to Nathalie Smuha, assistant professor at Leuven University, the rise of DeepSeek is an opportunity for LLM developers in Europe and around the world to challenge themselves to find ways to be more innovative “without being the ‘biggest’ one”.
“It shows that the US model of always throwing more money and data at it is not the only way. And it is also a reminder that regulation – such as a restriction on the export of chips – does not need to stand in the way of innovation,” she told Euronews.
Lucie-Aimée Kaffee, EU policy lead at Franco-American AI company Hugging Face – which provides access to a code library for evaluating machine learning models and data sets — echoed these statements.
“By focusing on responsible, high-quality AI rather than pure scale, European developers have a unique opportunity to play to their strengths: advancing AI that is efficient, trustworthy, and tailored to diverse applications,” she said.
AI continent
Henna Virkkunen, European Commissioner-designate for Tech Sovereignty, Security and Democracy, pledged to make Europe the ‘AI continent’, but the bloc trailsbehind other regions in the world when it comes to investment in AI.
A reportpublished in May 2024 by the EU Court of Auditors warned that Europe has so far had little success in developing its own artificial intelligence ecosystem, and has failed to accelerate investment on a par with global leaders.
The overall AI investment gap between the US and the EU more than doubled between 2018 and 2020, with the EU trailing by more than €10 billion and newly elected US President Donald Trump just announced billions of dollars in AI investments.
Claes de Vreese, University Professor of Artificial Intelligence and Society at the University of Amsterdam, told Euronews that with the rise of a Chinese LLM player, Europe is now caught in the middle of the US and Chinese developments.
“Rather than entering into a race about being faster or bigger, the EU, also in the light of the Draghi report, needs to come up with an innovative model that explicitly centers data protection and security. Consumers will want this in the long run, so this can turn out to be a viable, responsible, and competitive strategy,” De Vreese added.
Paul De Hert, Professor at the Vrije Universiteit Brussel (VUB), said yesterday during a data privacy conference co-organized by the Council of Europe in Brussels that the open source technology behind the DeepSeek development is the most interesting.
De Hert explained that by democratising this technology, it challenges the current power imbalance between countries and companies with almost unlimited resources compared to others – and that Europe could benefit from it.
Data protection issues
So far, AI LLMs have clashed with Europe’s regulators about the use of data to train their systems.
Last August, the Irish Data Protection Authority fileda complaint at the country’s High Court, over the use by X’s Grok of personal data to train its artificial intelligence tools.
Anna Rogers, Associate Professor at the IT University of Copenhagen, said that the rising public awareness of privacy issues might lead to more opportunities for “privacy-first, socially sustainable EU-based businesses.”
“The waning public trust in the tech giants could be an opportunity for Europe, as it is widely known to offer so far the strongest privacy principles on the globe,” she said.
Hugging Face’s Kaffee said that EU companies could become more successful if AI models developed outside Europe face regulatory restrictions, particularly regarding data usage and compliance with EU laws.
“As scrutiny increases, AI companies operating outside Europe may struggle to deploy their models within the EU market, especially for high-risk applications,” she said.