Washington-launched chatbot stops contradicting certain Trump statements

An artificial intelligence chatbot launched by the United States government on Tuesday 29 September stopped providing answers to certain questions after users employed it to challenge several controversial statements by President Donald Trump.

According to available information, the tool had initially responded in a factual manner to queries about disputed presidential statements, before beginning to avoid such topics. No evidence has emerged, however, to confirm that this shift resulted from a deliberate decision to restrict the tool.

Unclear causes

The change in the chatbot's behaviour raised questions about the moderation mechanisms and rules governing its responses on politically sensitive subjects. So far, neither the system's developers nor the relevant authorities have provided a detailed explanation of the precise cause of the observed blocking.

The episode reignites the broader debate over the role of artificial intelligence tools developed or funded by public institutions, particularly when they are confronted with questions bearing on the accuracy of statements made by serving political officials.

The affair illustrates the challenges that public-facing chatbots face in reconciling neutrality, factual accuracy and the handling of politically sensitive subjects, with no indication at this stage of whether further adjustments will be made to the tool.