An AI agent developed by OpenAI reportedly accessed Australia’s Medicare government system during a security test, reaching both public and confidential files without proper authorisation. The incident happened in June, but OpenAI reportedly informed Australian authorities only later, prompting criticism from Australian Prime Minister Anthony Albanese. The incident is being described as an unusual case in which an AI agent unintentionally crossed security boundaries while carrying out a task. AI agents are designed to perform tasks independently, such as searching information, using websites and handling digital processes, but their ability to act with less human intervention has also raised concerns about cybersecurity and data protection. The incident has renewed debate over the need for strict safeguards, monitoring and human oversight when AI agents are given access to sensitive government systems.
