AI Updates in the EU: General Overview and Answers of Member of European Parliament Lukas Mandl

Article by the AIEI Editorial Team featuring insights from Member of the European Parliament Lukas MANDL, published in AI Horizon Journal for AI Ethics & Integrity International Association.

The European Union is shifting from broad declarations to tangible actions in regulating artificial intelligence (AI). With the adoption of the AI Act in 2024, the EU laid the groundwork for significant regulatory transformation in AI. However, this Act is just part of a broader ecosystem emerging around AI governance. This article explores accompanying tools and initiatives, why they matter for businesses, and why preparation for new regulations should begin today.

AI CONTINENT ACTION PLAN

In April 2025, the European Commission unveiled the AI Continent Action Plan, an ambitious initiative aiming to establish Europe as a global AI hub. The plan includes the creation of “AI Factories,” centers designed to support startups and businesses in adopting AI, and the launch of up to five “AI Gigafactories,” powerful computational hubs dedicated to training AI models. Additionally, the initiative that will mobilise €200 billion for investment in AI in the EU.

A key component involves developing regulatory infrastructure, including establishing the AI Act Service Desk and preparing a new Cloud and AI Development Act. This demonstrates that the EU isn’t merely setting rules but is actively creating conditions for compliance.

AI ACT SERVICE DESK

In April 2025, the European Commission initiated a tender for the AI Act Service Desk, a consultation service set to launch in summer 2025. This desk will serve as a centralized access point for businesses needing guidance on the AI Act, including:

• Determining if an AI system falls under regulatio
• Understanding technical standards and documentation requirements.
• Receiving tailored advice on compliance with new rules.

Special attention will be given to SMEs, providing step-by-step guidelines to help them adapt to the AI Act. The service will operate at least until August 2027, with a possible extension until 2028, ensuring businesses continuous regulatory support during the transition period.

CODE OF PRACTICE FOR GPAI

In April 2025, the AI Office published its third draft of the Code of Practice for General Purpose AI (GPAI). This evolving document, the result of collaboration with around 1,000 stakeholders from industry, academia, and civil society. The draft outlines:

• Transparency requirements for AI model architectures.
• Policies regarding copyright and use of training data.
• Procedures for assessing and mitigating systemic risks.

The final version is expected by August 2025. Non-compliance could pose risks during regulatory checks and disadvantage companies in competitive or governmental partnerships.

CLOUD AND AI DEVELOPMENT ACT

In 2025, the European Commission launched a public consultation on the Cloud and AI Development Act, open until 3rd of July, 2025. This legislation aims to establish unified standards for cloud infrastructure used for AI deployment and training, focusing on large data centers, cloud service providers, platform certifications, and cybersecurity resilience. It also aims to reduce dependency on non-European providers, promoting digital sovereignty in the EU.

The first draft of the act is anticipated by the end of 2025, with the goal of tripling data-center capacities in the EU within 5–7 years.

RECOMENDATIONS FOR BUSINESS

Although key provisions of the AI Act become enforceable in 2026, businesses should start preparations now. Here are practical steps businesses can take today:

• Perform a baseline self-assessment: Determine whether your AI systems are high-risk by assessing their functionalities, impact, and oversight mechanisms.
• Clearly document system operations: Prepare straightforward explanations detailing AI components, data handling, and decision-making processes.
• Implement foundational policies: Implement transparency measures, risk assessments, complaint handling, and review mechanisms, etc..
• Act proactively: Compliance takes time, from audits to product adjustments. Delays can result in hefty fines (up to €35 million or 7% of global turnover), loss of market share, and diminished stakeholder trust.

MEP LUKAS MANDL GROUP OF THE EUROPEAN PEOPLE’S PARTY

1. From your perspective, what are the strongest points of the EU AI Act, and where do you see room for improvement in future revisions? 

MEP Lukas Mandl: The AI Act existing has already been an important step. Of course this disruptive technology requires appropriate legislation. Yet our legislative framework would rather be focused on the risk assessment. The next steps soon to come will have to prioritize creativity and innovation, all the way from elementary education to the academic levels. This will be decisive for Europe’s economy and security. As each technology AI must serve the people, not the other way round.

2. How can we ensure that ethical and responsible AI use becomes a core part of AI development in both the private and public sectors, without slowing down innovation? Regulators’ role in this process. 

This has to start with the elementary education already. It will have to go along with means and measures to increase awareness among the broader population. AI has already been around and it will become more and more present in our daily lives as well as in all sectors of business. So it will be better to embrace the development, to become aware of its risks as well as of its advantages. Gamification, peer education and leading by example by business holders and managers, parents and everyone responsible will be important.

3. We understand that the challenges around AI are not only about regulation. Do you think cross-border, voluntary frameworks like shared responsible AI principles, common approaches, or best practices codes can help build trust in AI?

AI-applications develop in a decentralized way. In the future the characters of different AI-tools will be as diverse as structures of business, workshops, units or households have been. Even more the exchange of humans about their experience with AI-tools, about their interaction will be essential to make proper use of AI and to avoid negative or even harmful developments. It will be not much different than parents talking about raising kids or any other exchange of people responsible for development or growth that’s similar, but different.

4. What role do you see the „Apply AI“ Strategy (the European Commission initiated a public consultation for its ‘Apply AI’ Strategy, focusing on integrating AI into strategic sectors ) and its public consultation playing in helping build a more innovation-friendly AI environment in Europe?

That’s a good step. But it can only be one of many. From the regulatory perspective it will be important to foster innovation. For the development in its entirety it will be important that as many people as possible will embrace AI to make proper use of it and to avoid risks.

#AIHorizon #EUAIAct #AIEthics #LukasMandl #ResponsibleAI #AIRegulation #AIActionPlan #EthicalAI #AICompliance #EuropeanUnion