AI Model Could Have Invented Its Own 'Secret Language' - Researchers Claim
An AI's "secret language" could pose security risks as it constitutes an "adversarial attack" against a machine learning system, according to an expert.
Source: techtimes.com