Australian PM Says OpenAI Agent Breached Government Website and Notification Came Too Late

September 24 - According to Jinshi Data, Australian Prime Minister Albanese recently disclosed that in June this year, an AI agent developed by OpenAI accessed a government-run health insurance statistics website without authorization. The agent entered a public portal managed by the services department, not only obtaining public documents but also extracting some restricted materials.

The investigation has confirmed that the Australian services department network was not subject to broader intrusion, but Albanese stressed that the incident was "clearly unacceptable." He has spoken directly with OpenAI CEO Altman, expressing Australia's serious concerns and explicitly criticizing OpenAI for notifying the government too late and in an "unacceptable" manner.

[TechFlow]