Ethical Application of Generative AI in the EU : extending Deloitte’s Trustworthy AI Framework for modern AI capabilities whilst adhering to the EU AI Act.
Piccirilo, Felipe (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025051812946
https://urn.fi/URN:NBN:fi:amk-2025051812946
Tiivistelmä
This thesis investigates how ethical frameworks are impacted by AI’s LLM advancements (Large Language Models, also known as Generative AI), specifically on the example of Deloitte’s Trustworthy AI(TM) framework. Despite being a robust tool, Deloitte’s framework contemplates AI capabilities preceding LLMs and in a different context, thus it became somewhat obsolete. This research proposes an enhanced Ethical AI Framework, augmenting and fixing gaps of the existing one, contextualizing its application within the European region, and considering the European Union AI Act, which intends to regulate and mitigate risks with AI applications.
The research was conducted utilizing qualitative analysis of existing ethical and responsible AI frameworks globally, and discussions with Deloitte’s expert on AI implementation. Client projects analysis and experiences were used as part of the investigative process, helping to identify gaps and proposed solutions.
The thesis is composed of an elementary introduction on Ethics, AI, and frameworks for ethical implementation of AI. It also describes in detail the current state analysis of the existing framework and how effective and defective it is. The EU AI Act structure is discussed and how it converges or diverges from the ethical framework central to this thesis.
The outcome of this thesis is an enhanced Ethical AI Framework contemplating LLMs in recent geopolitical considerations, a practical list of steps to implement abstract ethical concepts into practice, and a case study of its application taking into consideration the EU AI Act regulation. The result can help private and public sectors to harness benefits of early use of AI acting ethically and responsibly while waiting for the ever-evolving regulatory landscape. It contributes to a constant discussion about purpose of new technologies to humanity and social impact in the European Union.
The research was conducted utilizing qualitative analysis of existing ethical and responsible AI frameworks globally, and discussions with Deloitte’s expert on AI implementation. Client projects analysis and experiences were used as part of the investigative process, helping to identify gaps and proposed solutions.
The thesis is composed of an elementary introduction on Ethics, AI, and frameworks for ethical implementation of AI. It also describes in detail the current state analysis of the existing framework and how effective and defective it is. The EU AI Act structure is discussed and how it converges or diverges from the ethical framework central to this thesis.
The outcome of this thesis is an enhanced Ethical AI Framework contemplating LLMs in recent geopolitical considerations, a practical list of steps to implement abstract ethical concepts into practice, and a case study of its application taking into consideration the EU AI Act regulation. The result can help private and public sectors to harness benefits of early use of AI acting ethically and responsibly while waiting for the ever-evolving regulatory landscape. It contributes to a constant discussion about purpose of new technologies to humanity and social impact in the European Union.