Robo Jetten real: BNNVARA reveals AI text generator behind the prime minister

According to recent reports, the Dutch prime minister, Rob Jetten, is using artificial intelligence (AI) to draft his speeches and messages. This involves creating draft texts using a generative language model based on vast amounts of sample data. These texts are then edited and verified by human editors, meaning that there is no complete automation in the publication process. The revelation has sparked a robust debate about the transparency and reliability of government communication, which are closely linked to the European AI regulation.

The Government Information Service (RVD) has deployed these generative AI tools to increase the speed and consistency of communication. This approach helps formulate core messages and responses to frequently asked questions more efficiently, while editors critically assess the final texts. However, the works must comply with guidelines regarding transparency, especially since the European AI Act requires that the use of AI in public communication is clearly stated.

In addition to the obligation for transparency, there have been calls from the House of Representatives and the press for the verifiability of AI-generated texts. This can be supported by keeping logs of versions and sources, which is also necessary for oversight and auditing. While generative AI offers opportunities to accelerate writing tasks and maintain consistency in tone and style, it also carries risks. The possibility of errors and inaccuracies in the generated content, as well as the chance of a too generic writing style, require careful final editing and source verification.

Moreover, the General Data Protection Regulation (GDPR) is relevant for government communication using AI. This leads to attention for data minimization and the need for a secure working environment. For structural deployment, Data Protection Impact Assessments (DPIAs) are recommended to identify risks and implement measures. The Dutch Data Protection Authority can oversee compliance with the requirements set in supplier contracts. Transparency towards citizens is essential, so that people know when AI has been used in text production.

To streamline the use of AI in government communication, there is a call for a public framework that provides guidelines for labeling, logging, and fact-checking. This can help prepare for the phased implementation of the AI regulation by 2026 and ensures that an open government maintains control over communication.

Read the full article from AI Insider.