A recent survey of over 100 high-tech AI startups which drive AI innovation by developing diverse AI systems, with both specific and generic purposes, and 15 Venture Capital firms (VCs) in the EU, has shed some light on potential startup challenges stemming from the EU’s AI Act proposal. This survey conducted by a coalition of different institutions which focus on AI stipulates that: 

• 73% of the surveyed VCs expect that the AI Act will reduce or significantly reduce the competitiveness of European startups in AI.
• About 33% – 50% of the surveyed startups would classify their AI Systems as high-risk under the current classification, and therefore will be subject to the EU AI Act’s obligations. This goes way beyond what the European Commission assumed in its Impact Assessment of the AI Act (5-15%).
• General Purpose AI (GPAI) obligations are written with large technological businesses  in mind, but many European startups will inevitably fall in the scope. More specifically, 45% of the surveyed startups consider their AI System to be General Purpose AI (GPAI henceforth), making them subject to the obligations of GPAI providers. The impact of this development has not been assessed, as GPAI was neither considered by the 2021-draft AI Act developed by the European Commission nor by the accompanying Impact Assessment.

Startups that are part of AI value chains not only need to meet the general AI Act obligations, but also to set up contractual arrangements to exchange relevant information with their customers, creating additional burden and complexity.

• In light of these developments, 16% of AI startup founders will consider halting AI development or relocation outside the EU. Notwithstanding,  the EU AI Act affects AI Systems which are placed on or put into service in the EU market and hence is not limited to organisations located in the EU. 
• 50% of startups surveyed are concerned that the AI Act will slow down AI innovation in Europe.
• For high-risk AI Systems (and their providers), the additional requirements and obligations are expected to be a significant challenge for the surveyed startups in terms of technical and organisational complexity and compliance cost. Around half of the startups that participated in the survey expect compliance costs to be on the level expected by the EU (i.e. €160.000-330.000) but 19% of them expect the costs to be higher.
• Reallocation of VC funds: low-risk Vs high-risk.

Most of the VCs expect the AI Act to reduce the competitiveness of European AI startups and to cause a VC investments shift towards AI Systems with a specific purpose, in low-risk applications, and, to some extent, to non-AI startups and outside of Europe. 

Thus, startups developing high-risk AI systems are less attractive to investors and at the same time even low-risk AI startups are affected by the proposed legislation. The survey shows that just the risk of an AI system developed by a startup being classified as high-risk or GPAI creates uncertainty for startups. VCs are drawn to certainty and will inevitably choose to invest in markets outside of the EU. Consequently, investment will drop and job creation by AI startups will come to a halt. A solid rationale for the low-risk classification needs to be in place, specifically if the Use Case falls into the category of GPAI. 

Policy Recommendations

1. Keep European competitiveness at the centre of the discussion.
2. Reduce the amount of high-risk use cases: Narrow the high-risk criteria to get closer to the anticipated 5%-15% of affected AI systems.
3. GPAI: Consider the role of startups as GPAI providers in light of GPAI-related obligations.
4. Foresee bottlenecks and systematically debottleneck the AI Act’s obligations e.g. in the area of third party conformity assessments to not additionally slow down innovation.
5. Conceptualise regulatory sandboxes as drivers for innovation in a protected, but attractive environment.
6. Update the coordinated plan and take the needs of European AI startups into account, specifically in areas that are considered very difficult or costly. Reduce cost of compliance.