The EU AI Act, which recently became law, has raised concerns among developers that it may hinder innovation and competition from open-source models and startups due to potential regulatory capture by large private labs. The Act categorizes AI applications into four risk levels: prohibited, high, limited, and minimal, with compliance obligations varying accordingly. Most developers will not be affected unless their applications fall into the high-risk category or involve training foundation models exceeding 10^25 floating point operations. High-risk applications require comprehensive compliance measures, including risk management, data governance, and transparency. Open-source model providers must adhere to specific requirements, such as summarizing training data and adhering to the Copyright Directive. The EU AI Act aims to balance regulation with innovation, despite concerns that its definition of systemic AI systems based on compute thresholds might be arbitrary. Humanloop, a platform for AI compliance, offers tools to help developers meet these new regulatory requirements.