Researchers have developed a new technology that relies on integrating invisible “digital watermarks” into texts processed by artificial intelligence, with the aim of enabling detection tools to identify automatically generated content with high accuracy, without affecting the quality of the text or the smoothness of its reading.
The new technology relies on introducing subtle and unnoticed modifications in the patterns of word choice and sentence formulation during text production. These patterns act as a hidden digital fingerprint that verification systems can capture immediately, making it easier to distinguish fabricated texts or processed reports through different artificial intelligence models.
This step comes in light of escalating international concerns about the spread of fake news, and the misuse of artificial intelligence tools in academic research and digital marketing, as experts are betting on this watermark to reduce digital misinformation and rebuild trust in content published online.
