ERINIA - Evaluating the Robustness of Non-Credible Text Identification by Anticipating Adversarial Actions
Within ERINIA, we analyse the emerging content filtration methods to see if they deliver on their promises. Can AI accurately detect non-credible text, such as fake news? Can human adversaries fool such automatic filters? What other harms could this technology bring about?