Update AI Act proposal of 10 July

Update AI Act proposal of 10 July

The future of artificial intelligence in Europe hangs in the balance as EU-legislators debate critical rules that will shape the approach on Artificial Intelligence (AI). Will Europe build an AI regulatory regime that stifles innovation with bureaucracy? Or can it function as a flexible framework that enables growth while addressing concerns?

In the latest proposal of the AI Act revealed earlier last month, it became clear that with compromises reached on numerous articles, the AI Act is steadily moving forward in negotiations. Though challenges remain, the negotiators appear committed to balance innovation in AI with managing risks through proportionate regulation.

The proposal uses a colour coding system that indicates the status of the articles. Green highlighted articles are provisionally agreed on between the EU institutions involved in the negotiations. These institutions are the European Parliament, the Council, and the Commission.

• Green suggests significant progress towards final agreement;
• Yellow indicates that final checks are needed, but the provisions are likely to be agreed on;
• Red means further discussion and compromises are required. These may be more controversial or complex provisions, that might not make it to the end.

The goal is to reach a common position between the three institutions. The provisional agreements must still be confirmed, but the colour coding gives a glimpse of where consensus is emerging.

Businesses which have been anxiously awaiting clarity on the assessment of AI systems in the EU may finally have reason for optimism, now that the green and yellow highlights can provide some perspective. Even if the details haven't yet been worked out.

There is a lot of pressure on finding the right balance between accountability and flexibility. Legislators seem to recognise the dangers of both burdensome bureaucracy that stifles innovation and lax oversight that enables abuse.

While conformity rules gain ground, proposed articles about regulatory sandboxes, that would enable controlled AI testing, still face scepticism and remain highlighted in red. In other words, these articles will be subject to later adaptation, depending on the resolution of overarching issues such as governance, basic models/general-purpose AI systems or law enforcement exemptions.

The journey ahead aims to harmonize innovation and ethics for a technologically empowered Europe. We will closely monitor developments regarding the AI Act. Would you like to know more about the proposal? Please feel free to contact us.