EU's AI Act: SMEs Must Assess, Document AI System Risks
The EU is set to introduce the AI Act, a comprehensive regulation impacting businesses using artificial intelligence. Small and medium-sized enterprises (SMEs) must assess and document their AI systems' risks, with high-risk systems facing significant obligations.
Philip Heider, an AI expert at heyData, advises companies to start by conducting a stocktaking of their AI systems and determining each system's risk class. High-risk AI systems, such as those used in medical products or critical infrastructure, require substantial documentation including training data and bias mitigation measures. Even limited-risk systems, like chatbots or spam filters, need transparency measures like labeling AI-generated content.
The AI Act distinguishes between different roles in the AI ecosystem, with most SMEs acting as users. Companies must document their risk assessment process, even with minimal risk. Failure to comply with the AI Act can result in fines of up to seven percent of annual turnover.
The AI Act is a significant development for companies using AI, with SMEs needing to assess and document their AI systems' risks. High-risk AI systems require substantial documentation and transparency measures. Companies must take action now to ensure compliance with the upcoming regulation.
