Dutch Prime Minister Rob Jetten has misinformed parliament about his use of artificial intelligence.
On September 29, answering written questions from Pepijn van Houwelingen of the right-wing Forum for Democracy (FvD), the prime minister wrote that “no use was made of AI” on his official accounts.
Though deeper research has found that five posts on those accounts, from the past six months, were written by it.
Newspaper De Volkskrant ran the official MinPres feed through Pangram, the detector it used in September.
Five posts, averaging about 200 words, came back as generated.
Researchers at the universities of Chicago and Maryland found Pangram wrongly flags about one in 10,000 human texts.
The paper also ran 700 posts published by Jetten and his party before ChatGPT was released in 2022 through the same model. None was flagged.
Two language experts told the paper that five misses at that length was not a serious possibility.
Marijke Van Vlasselaer, an AI researcher, said the posts shared “the same cadence and structure”, opening with “The Netherlands and Germany are neighbours”, “The world is restless” and “Everyone knows it”.
Van Vlasselaer warned that such outsourcing of language and thinking to mathematical models will break the trust between people.
Popular blogsite GeenStijl called the parliamentary answer “stone faced lying”.
The prime minister’s use of AI has been a talking point since the paper found, on September 10, that at least 232 posts on Jetten’s personal accounts and D66’s party feeds had been machine-written: 68 in his own name and 164 from the party.
From June 4, 2025, the day after the Schoof cabinet fell, about 30 per cent of Jetten and D66 posts were generated.
The subjects included sensitive issues such as Ukraine and the MH17 commemoration.
A page and a half of his book Hoe het wél kan (How It Can Be Done), and lines in three campaign videos, were flagged too.
Reacting to the debate, the Government Information Service (RVD) said the texts had not been taken over wholesale from a chatbot and drew on policy files, meeting notes and material used before, so the 2025 civil-service rules did not apply.
Those rules ban the wholesale adoption of machine text and require the use to be declared.
The service has since pointed to a looser rule on undisclosed AI use. It was drawn up on September 17, a week after the first story.
Jetten told RTL Nieuws: “You put something in, you take something out, and you always check it”.
Wim Voermans, professor of constitutional law at Leiden University, said the written answer had shut that door.
“If you state flatly that AI was not used, and it can be shown that it was, that cannot be explained away politically.”
The information service has since pointed to a looser rule on undisclosed AI use.
It was drawn up on September 17, a week after the first story.
