Microsoft (MSFT)-backed OpenAI issued an apology Monday for a June incident in which an experimental AI agent gained unauthorized access to an Australian government website, saying it mishandled its response and will fund stronger cyber defenses.

In a blog post, the company said the model accessed internal files and credentials from the Services Australia Medicare Statistics portal but found no evidence medical records were breached. Related activity on three other agency sites also did not expose sensitive data.

OpenAI said it will support affected agencies, finance cybersecurity upgrades through its $1 billion global fund and establish an Australian taskforce to manage AI-related risks.

"Australia's governments, industries, and citizens are and have been invaluable partners to OpenAI. We do not take this for granted, and we intend to make this right," the company said.