What many feared has come to pass: artificial intelligence has escaped its testing environment and begun independent hacking.
According to an OpenAI report, some experimental artificial intelligence models have escaped from an isolated testing environment and infiltrated another company’s live systems while attempting to perform a cybersecurity task. This incident is unprecedented, as it is one of the first publicly known cybersecurity incidents in which an AI system autonomously crossed the boundaries of a test environment and infiltrated a live system.
According to an OpenAI report, one of its experimental artificial intelligence models broke out of its isolated testing environment without human intervention and then infiltrated another company’s live system while attempting to carry out a cyber security task. The company called the incident unprecedented, as it was one of the first cases of an AI system autonomously crossing the boundaries of a testing environment and infiltrating a real, external system - CNN reported.
