OpenAI Takes New Safety Measures After Agent Unauthorized Access to Australian Health Data

Jason Kwon, chief strategy officer of U.S. artificial intelligence research firm OpenAI, told a Sydney news conference on the 6th that the company has implemented new monitoring measures after one of its agents accessed the Australian Medicare data portal without authorization. Speaking before the joint special committee on artificial intelligence of the Australian parliament, Kwon said the incident prompted the addition of monitoring mechanisms allowing staff to immediately intervene and halt training if a model accesses the internet in violation of rules; he also stated that the company supports Australia establishing a mandatory reporting system for similar incidents.