Joint Statement: The AI Act: Four recommendations for Member States from Europe's digital industry

The proposed AI Act will be the first legislation of its kind in the world. Europe’s tech future is at stake – this is an opportunity not to be missed.

DIGITALEUROPE and its national association members are supportive of the AI Act’s objectives and focus on high-risk use cases. AI will be the driver of innovation in the coming years, with some estimates foreseeing that in 2035, half of the growth of many European countries may be AI-driven.¹ However, only 3% of the world’s AI unicorns come from the EU² and we see ten times more private investment in AI in the US, and five times more in China.³ To fully benefit from AI, our focus must be to create a Europe where innovative companies can flourish.

To ensure uptake, trust in the technology is essential: the AI Act has a role to play to help business bring trustworthy AI solutions to the market. However, the AI Act is a complex legislative proposal, creating a new framework at the intersection of traditional ICT product compliance and the protection of fundamental rights. Implementation will be challenging, especially as businesses and market surveillance authorities will navigate through uncharted territory. Some estimates show that, for an SME of 50 employees, placing one AI-enabled product on the market could result in compliance costs of well over €300,000.⁴ We are currently conducting a sandboxing assessment of the expected impact on SMEs and startups.

The work undertaken by the Council under the Slovenian, French and Czech presidencies has helped to increase legal certainty for businesses. However, a few challenges remain.

As Member States prepare to adopt their General Approach, we encourage the Council, as part of the upcoming trilogue negotiations, to consider the following points:

Regulatory sandboxes are crucial to ensure proportionate obligations and smooth implementation. We recommend both pre-regulation sandboxing to test the law’s effects on businesses and post-regulation to help them prepare for compliance. Member States must commit to establishing such sandboxes, notably at a pan-European level.
Giving the right responsibilities to the right actor in the AI value chain is key. The AI Liability Directive will rely on the allocation of duties under the AI Act, which needs to be further refined, including recent developments on general-purpose AI. The AI Act should, when relevant, provide added flexibility by ensuring the freedom of the parties to allocate responsibilities through contractual obligations.
The AI Act may still conflict with existing legislation such as the Machinery or Medical Device regulations. Ensuring alignment and avoiding duplication of procedures would greatly simplify compliance in many sectors where Europe leads today and must remain competitive.
National authorities must cooperate to avoid a fragmented implementation. Appointing one central contact point per Member State and a strong EU coordination mechanism under the AI Board will ensure consistent application and enforcement.