The European Union has published guidance on what constitutes AI systems under the new AI Act. Last summer, a risk-based framework was implemented to regulate the application of artificial intelligence. The first compliance deadline (banned use cases) kicked last weekend.
Determining whether a particular software system falls within the ACT range is an important consideration for AI developers. This risks fines up to 7% of global annual sales for violations. The committee’s 13-page guidelines may be closely analyzed by companies.
That said, similar to the EU guidance on banned use published earlier this week, the advice is not binding. The committee also added that the guidance is “designed to evolve over time and will be updated as needed, particularly in light of practical experiences, new questions and use cases that arise.” Masu. Given how fast and paced the AI field is, the task of understanding how the law applies may remain a work in progress.
“An automatic decision or thorough listing of systems within or outside the definition of an AI system is not possible,” the EU warns in its advice document. Set expectations accordingly.