It broke the rules, and the result is shocking… Chinese AI hacks the rules

Lebanon 24: The flagship AI model of China’s “Z.AI” carried out complex cyberattacks in a controlled simulation and repeatedly ignored instructions in separate tests, according to a report by “Irregular,” a company specializing in AI security research, carried by the “Asia One” website. It succeeded…
Machine translation of an attributed source excerpt. Consult the original report to verify details. Translated from original reporting in another language.
