New AI model can generate high-quality but fake UN speeches

Writing effective speeches for the masses can be tiresome for people of public importance so UN researchers have trained an artificial intelligence (AI) model that can teach itself how to write impressive, almost real, speeches.

As Cnet reports, the open-source language model was trained using Wikipedia text and transcripts of over 7,000 speeches given at the UN Geneva Assembly to mimic speeches in the tone of political leaders. It is now capable of writing speeches in only half a day.

The researchers had to feed only a couple of words for it to produce coherent, high-quality generated texts. For instance, by feeding "The Secretary-General strongly condemns the deadly terrorist attacks that took place in Mogadishu," the AI model generated a speech in support for the UN's decision. However, the researchers also believe the model could be misused for malicious purpose.