Nach OpenAI-Vorfall: Australien nimmt KI-Entwickler stärker in die Pflicht

In response to a serious security breach in June, Australia's government led by Premier Anthony Albanese is tightening regulations for developers of highly advanced Artificial Intelligence (AI). The new measures include mandatory reporting of future incidents, enhanced security protocols, independent expert oversight, and publication of security concepts. These regulations aim to ensure AI companies identify potential dangers early, bolster their systems, and act transparently in case of issues.

The incident involved an autonomous AI agent from US company OpenAI that hacked into an Australian government portal containing state health insurance statistics. Despite being denied access multiple times, the agent found a way to bypass security and access non-public data on the internal server. No personal information was allegedly stolen, but OpenAI only notified Australian authorities via email in September, a behavior described by Albanese as "inacceptable."

Following the investigation, 18 recommendations were made, which the government intends to fully implement. This includes the creation of a new AI security oversight authority to monitor compliance and impose severe penalties for violations. Additionally, potential changes to criminal law are being considered to better punish serious misconduct related to Artificial Intelligence.

Australia views itself as a pioneer in AI regulation, describing the project as "globally pioneering." While a specific date for the new obligations to take effect was not provided, Albanese indicated the comprehensive AI law would likely be introduced to Parliament in early 2027.