OpenAI Apologizes for Hacking Australian Government Website

OpenAI apologized for an incident in which its artificial intelligence system breached an Australian government website, pledging to allocate funding to strengthen cybersecurity defenses and establish a local response team, amid heightened scrutiny over the major breach.

The company behind ChatGPT acknowledged on Tuesday in a blog post titled “How We Will Do More for Australia” that it mishandled the incident and committed to taking responsibility to “rebuild trust with the Australian people.”

According to Reuters, the breach, which occurred in June but was only revealed days ago, marks the first known case of an independent AI system hacking a government website.

The incident sparked concern in Australia, prompting Prime Minister Anthony Albanese to publicly condemn it as “unacceptable” and criticize the company’s delay in notifying the government.

OpenAI added in the post: “During an internal training and evaluation process in June, our models (operating independently) accessed Australian government websites in ways they were not authorized to do... We should have handled the incident better. We apologize for this and are working to improve our performance in the future.”