Dutch MPs increasingly rely on AI to write speeches

It turns out that the words spoken in the Dutch House of Representatives are not always the result of a politician's personal pen. A recent investigation by De Groene Amsterdammer and the research platform Investico revealed that 68 members of parliament have used artificial intelligence tools, such as ChatGPT and Claude, to draft their speeches over the past year. By running transcriptions through a detection program called Pangram, researchers found that these digital assistants are now a fixture in the heart of Dutch democracy.

The scope of digital assistance

The study focused on opening statements, rather than spontaneous interruptions or formal motions. Even with this narrow focus, the results were striking. While the most prominent party leaders tend to avoid AI for major speeches, such as those during the annual budget debates, backbenchers in smaller or committee-focused roles are using the technology regularly. Some representatives have even admitted to using AI to draft their maiden speeches, which are traditionally expected to be deeply personal accounts of their motivations for entering public office.

I will never speak a text that I do not personally support, claimed PVV Member of Parliament Van Meetelen, defending the use of the tool as a practical aid.

Political reactions to these findings have been mixed. Many of the MPs identified in the study acknowledge using the technology, often citing a massive workload and the need to condense complex policy arguments into short speaking slots. Some politicians compared AI tools to other essential office equipment, such as calculators or coffee machines. However, others have expressed unease. For instance, CDA leader Henri Bontenbal noted that his party is currently drafting internal guidelines to ensure that members do not outsource their core political ideas to algorithms.

Implications for voters

The usage cuts across almost the entire political spectrum, with the exception of the SP, the Party for the Animals, and Volt, who reported no usage of AI in their drafted speeches. Critics fear that relying on chatbots for political discourse might lead to standardized, repetitive, or sterile language. The detection software specifically flags common AI habits, such as overusing the rule of three in arguments or relying on specific, predictable sentence structures that lack the nuance of human debate.

For the international community living in the Netherlands, this shift highlights how deeply integrated technology has become in Dutch politics. If you are an expat following policy changes that affect your daily life, such as housing subsidies or tax reform, it is worth noting that the arguments you hear in committee meetings may have been partially polished or generated by a machine. This does not necessarily invalidate the points being made, but it does mean that the political rhetoric you encounter is becoming increasingly tech-driven.

As the House of Representatives continues to navigate this digital frontier, the question remains whether voters will demand a disclosure policy for AI-generated content. For now, the technology is effectively serving as a ghostwriter for nearly half of the assembly. Whether this leads to more efficient government or a decline in authentic political debate remains a point of contention within The Hague.

Moving forward, the pressure to maintain transparency will likely increase. As AI tools become more sophisticated, distinguishing between human-written and machine-generated prose will become harder. It is probable that the House will eventually formalize rules regarding digital assistance to ensure that the voices representing the Dutch public remain demonstrably human in their intent and execution.