Last February, the European Commission published its White Paper on Artificial Intelligence announcing an AI regulatory framework to enable the safe development and deployment of AI in the EU. The White Paper release was coupled with the Commission’s consultation for stakeholders to express their opinion.

The European Commission wants to propose a risk based approach to ensure a proportionate regulatory intervention, with a distinction between high risk and low risk AI applications. Low risk AI could be subject to voluntary labelling and high risk AI could be subject to ex-ante and ex-post conformity assessments. Assessing the risks of an AI application will be based on the sector they belong to (such as medical, entertainment or education) and its intended use. The criteria employed to make this distinction are the public’s exposure to harm to legal values like life, health and property.

While a risk-based approach takes different applications of AI into account, we wonder what it will mean in practice for entrepreneurs innovating in high risk sectors such as health. Indeed, grey areas are likely to arise in defining if an AI application is high risk or low risk.

The ambition of AI legislation should be that even the startups that operate in high-risk areas should have a clear perspective to scale-up. We believe that new legislative requirements should be understandable and implementable for startup entrepreneurs. If ex-ante conformity requirements are introduced, how long would it take for an EU body to evaluate whether the application can enter the market or not? Building on the example of the Medical Device Regulation, where it can take 6 weeks to get approval, it could make it more complicated and lengthy for an entrepreneur to be able to put their product on the market. It might end up in the startup not being able to launch in the European Union and preventing EU citizens to benefit from a groundbreaking new product or service.

Assessment procedures and bodies should be built and adapted to the startup ecosystem. For innovation to thrive in Europe, the European Union should equip any new body responsible for AI labelling with the resources to match entrepreneur’s fast and dynamic working models!

Startups are all about disruption and unconvention. Hence, an adapted approach that does not apply the same blueprint to every startup, can be a first step to ensuring that regulation doesn’t stand in the way of innovation. A distinction between high and low risk sectors or use-cases should be designed while keeping in mind that there are ambitious startups operating in all fields. Without this approach, the European Union will not incentivise entrepreneurs to innovate in high risk AI applications within the European Union.

Startups have shown their potential in helping the recovery of Europe COVID-19 crisis with the thousands of services they’re offering in education, health or financial services. Entrepreneurs are consumers first, they innovate where they see a possibility to offer a better choice to the consumers. EU Policymakers should take this into consideration when drafting the AI regulatory approach.