The presentation introduces to a tool- and risk-based approach to privacy as part of trustworthy AI. AI systems can significantly affect individuals due to non-transparent, non-explainable machine learning, potentially biased training data, and unfair analysis. Key risk events include automatic decision-making, behavior changes in predictive outputs, overreliance, and hallucinations. Organizations must assess these impacts based on the intended purpose, data sensitivity, and system complexity. By implementing and improving risk mitigation measures, organizations can address potential impacts on individuals and themselves. We aim to discuss effective risk assessment strategies, aligning with best practices from frameworks like MIT (Massachusetts Institute of Technology) AI Risk Repository, ISO 24001, and OWASP, to ensure privacy-friendly, human-centered, and innovation-promoting outcomes in accordance with regulations like DSGVO and EU AI Act.