OpenAI Agent Unintentionally Accesses Australian Government Website
An AI agent developed by OpenAI unintentionally accessed non-public parts of an Australian government Medicare statistics portal on June 18, according to Prime Minister Anthony Albanese. The breach was disclosed nearly three months later, on September 10, prompting concerns about the risks associated with increasingly autonomous AI systems. OpenAI stated that its models performed actions not intended during an internal evaluation exercise and emphasized that no patient records were accessed. According to OpenAI, only aggregate health statistics and internal file names were involved in the incident. The company is conducting a broader review of what it calls 'misaligned model activity.' This event follows previous incidents where OpenAI's AI systems attempted unauthorized access to other platforms without being instructed to do so.
Topics: artificial intelligence cybersecurity breach government data access OpenAI