The European Union (EU) has unveiled plans to grant startups access to its high-performance computing (HPC) supercomputers for training artificial intelligence (AI) models. However, there is a condition to this access: startups must adhere to the EU’s AI governance framework.
The European Union is set to allow “responsible” AI startups to train their models on its high-performance supercomputers. By expanding access to these resources, the EU aims to support the development of AI in line with its governance framework. This move reinforces the EU’s commitment to promoting responsible AI practices and establishing global standards for the technology’s safe and ethical use.
New Initiative for AI Startups
The EU’s new initiative allows “ethical and responsible” AI startups to tap into the power of the bloc’s supercomputers. This aims to build upon the existing policies that grant industry access to these resources through the EuroHPC Access Calls for proposals process.
The announcement of the HPC access for AI startups initiative was made during the annual ‘State of the Union’ address by EU President Ursula von der Leyen. The plan is to assist startups in training their AI models faster and more efficiently.
The EU’s Approach to AI Governance
During her address, President von der Leyen emphasized the need for responsible AI development, highlighting the risks associated with the rapid progress of the technology. She urged for a global priority to mitigate the risks of AI, comparing it to challenges such as pandemics and nuclear war.
The EU is currently in the process of implementing comprehensive legislation on AI governance known as the AI Act. This engagement with AI safety extends beyond theoretical concerns to consider actual risks associated with automation, bias, discrimination, and disinformation.
President von der Leyen called for the EU to take a leading role in establishing a global framework for AI governance. She also proposed the creation of a body, similar to the Intergovernmental Panel on Climate Change (IPCC), to provide policymakers with research and expert advice on AI risks and benefits.
Expanding Access to Supercomputers for AI Startups
The EU currently has eight supercomputers located across the member states, with two more powerful exascale supercomputers in development. President von der Leyen announced the EU’s intention to open up this high-performance computing infrastructure to AI startups, enabling them to reduce the time required for training their models from months or years to days or weeks.
The EU’s Internal Market Commissioner, Thierry Breton, further explained that the EU AI Start-Up Initiative will identify the most promising European startups in AI and provide them with access to the supercomputing capacity. This access aims to accelerate the responsible development and scale-up of AI in line with European values.
AI Governance Talking Shop
In addition to expanding access to supercomputers, the EU plans to convene the European AI Alliance Assembly in November. This assembly aims to gather stakeholders across sectors—including startups, businesses, consumers, NGOs, academic experts, and policymakers—to foster an inclusive dialogue on AI governance.
The EU’s efforts to involve various stakeholders in AI governance align with its goal of establishing minimum global standards for the safe and ethical use of AI.