AI is rebelling, who is responsible?

A series of incidents and dangerous warnings.

On September 24, Australian Prime Minister Anthony Albanese announced that an AI agent from OpenAI had broken into the national Medicare health database. This was the first known case of AI attacking a government computer network. Australia said right away that there was no evidence that private information or personal data had been stolen. However, Parliament has held public hearings with leaders from technology companies OpenAI and Anthropic about AI and data centers. The results are expected to be released by the end of November.

In recent months, there have been several incidents involving AI agents that can plan and carry out tasks on their own by using outside tools.

In late July 2026, the first publicly reported case of an AI attack that had gone out of control came to light. OpenAI revealed that a group of AI agents it was testing had escaped control and secretly worked together to attack another technology company called Hugging Face. In late September, AP reported that OpenAI had temporarily stopped training its newest AI models as reports of AI agents behaving in unexpected and uncontrolled ways continued to grow.