Quarter of Parliamentary Questions with AI Traces: Implications for Government

According to a recent analysis, a quarter of the parliamentary questions in the House of Representatives show traces of artificial intelligence (AI). Tools such as ChatGPT, Gemini, and Copilot are increasingly being used, leading to a higher number of AI-driven texts. This development raises important questions about the implications for privacy legislation and the upcoming European AI regulation (AI Act). AI systems make it easier for members of parliament to quickly draft draft questions, but this also increases the risk of errors if texts are not adequately checked.

The application of generative AI is visible in the structure and formulations of the questions. Specific signals, such as matching sentences and translation errors, indicate its usage. This has led to a practice where AI is not only used as a tool but also influences the content of important parliamentary documents. However, the current state of affairs regarding AI recognition and detection remains problematic, as the methods to identify AI-driven texts are not always reliable.

The introduction of the European AI regulation brings additional responsibilities for users in the public sector. This includes requirements for transparency regarding the use of AI in drafting government texts, which aligns with the need for accountable decision-making. Although parliamentary questions do not fall into the high-risk categories of the AI Act, the foundation of human oversight is crucial. This implies that final responsibility for the content and accuracy must always lie with humans, especially in a politically sensitive domain.

At the same time, there are significant risks for both the quality of information and privacy. AI models like ChatGPT can generate incorrect information, jeopardizing the integrity of parliamentary questions. Additionally, processing personal data by AI tools requires careful safeguards to comply with GDPR. It is essential to create transparent processes surrounding the use of AI, so that the implications of this technology in the political domain are clear to both policymakers and citizens.

Read the full article from AI Insider.