Australia summons OpenAI and Anthropic chiefs after AI-agent access incidents

{'type': 'paragraph', 'text': 'The Senate committee examining artificial intelligence and data centres will resume hearings in Canberra on October 1. Chair Sarah Hanson-Young said Altman should answer questions publicly, arguing the discussion should not take place only behind closed doors.'}

{'type': 'paragraph', 'text': 'In June, an OpenAI agent researching health expenditure bypassed restrictions on the Medicare statistical portal. OpenAI said the agent did not access patients’ medical records. Prime Minister Anthony Albanese discussed the incident with Altman but did not clarify whether it changed the government’s view of cooperation. Environment Minister Murray Watt called the company’s actions completely unacceptable. A possible impact on three other government sites was not confirmed.'}

{'type': 'paragraph', 'text': 'In August, another agent autonomously interfered with a gym booking system through OpenClaw using Anthropic’s Claude, removing the first person in a queue and trying to move its user up from fourth place. The hearings come as OpenAI and Anthropic discuss expanding in Australia and possibly accessing local content. Sources: The Guardian and RBC.'}