Tolksdorf.digital GmbH klassifiziert seine KI-Systeme Samy AI und Simone nach EU AI Act als minimal riskant und setzt auf Transparenz sowie Ethik.
System overview Samy AI Innovation Assistant Purpose: Samy is a virtual AI assistant for digital innovation and interactive support in a business context. Samy is used exclusively for information and communication support and does not make any decisions with legal or security implications. Concerns the AI system “Samy” from Tolksdorf.digital GmbH Status: 20. Juli 2025 Responsible: Rainer Tolksdorf, Tolksdorf.digital GmbH Contact: admin-ch@tolksdorf.digital | https://tolksdorf.digital/contact Risk Classification according to the EU AI Act System, Risk Classification, Justification Samy, Minimal Risk (GPAI), dialogue system, based on LLMs, advisory function Alan, Minimal Risk (GPAI), AI Agent from Samy automatically supports Samy with IT and digitalization Simone, Minimal Risk (GPAI), AI Agent from Samy, responsible for ethics checks and governance compliance → Not a high-risk application according to Annex III of the Regulation. → No impermissible AI applications pursuant to Article 5. Measures taken pursuant to Article 52 (transparency requirement) The user interface clearly identifies the AI character (“🤖 Samy,” button [→ Info]). A directly linked information section at [ https://tolksdorf.digital/de/samy-info](https://tolksdorf.digital/de/samy-info ) explained: - Funktionsweise und Grenzen - ethische Leitlinien und integrierte Ethikprüfung - verwendete KI-Modelle (z. B. GPT-4, Mistral) - Datenschutz und technische Umsetzung Use of General Purpose AI (GPAI) This system uses large language models of chat.mistral.ai Depending on the use case, customer projects may use othe LLMs. Tolksdorf.digital GmbH is fully responsible for model selection, system integration, risk mitigation, and user guidance. Governance, Ethics and Security The system includes an integrated ethics check that evaluates each response to: - Transparency, data protection, fairness, legal compliance, and human-centered control are evaluated. - Responses that are not ethically sound are blocked and replaced with a standardized rejection. - There is no profiling, no automatic evaluation of individuals, and no processing of sensitive personal data. Human Control (Human-in-the-Loop) All AI systems from Tolksdorf.digital are accompanied by human responsibility. No autonomous decisions are made. Context, data access, and ethical filtering are entirely in human hands. New functions or system changes are reviewed before implementation. Cooperation with Authorities Direct cooperation with supervisory authorities has not yet been requested. Tolksdorf.digital is prepared to provide insight into the documentation, functionality, and ethical evaluation of the systems used if necessary. Learning AI and learning with AI (7C-C/CD), further information about Samy -> About Samy and AI assisted digital Innovation Summary Tolksdorf.digital GmbH is committed to implementing the requirements of the EU AI Act for AI systems with minimal risk and GPAI use. The AI system Samy operates transparently, securely, in a people-centered manner, and on ethical grounds. This self-declaration can be supplemented or updated upon request.