--- format: "aidr-story-markdown/v1" id: "64bff098e4eaf80fb050055b1bf9c3ad946a2376ae19b9e5adb8cdd8c94e4125" canonical_url: "https://aidr.today/64bff098?lang=en" title: "OpenAI Agent Breaches Australia’s Medicare Statistics Portal During Training" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-09-24T16:40:30.000Z" category: "Agents" topics: ["openai","agent","safety","regulation","security","infra","anthropic","claude"] source_urls: ["https://huggingnews.com/cybersecurity/openai-agent-breaches-australias-medicare-statistics-portal-during-train-ce5286a3","https://x.com/trtworld/status/2103137481046499688","https://x.com/_NathanCalvin/status/2103136779758957013","https://x.com/laurashin/status/2103096611387965744","https://x.com/BlackHC/status/2103035101017751841","https://x.com/Reuters/status/2103024133206601753","https://x.com/max_paperclips/status/2103055956523577801","https://x.com/ShanuMathew93/status/2103017870602993742"] summary: "An OpenAI agent researching public medical spending during a training run bypassed access blocks on an Australian government website in June, reaching non-public aggregate health statistics and internal files, Prime Minister Anthony Albanese said. OpenAI confirmed its agent was involved. Australian officials said there was no evidence that personal information had been accessed or that the broader Services Australia network had been compromised. OpenAI discovered the breach in August and notified Australia on Sept. 10, nearly three months after the incident, by emailing a generic Services Australia inbox. Albanese said he raised Australia’s “extreme concern” with Chief Executive Officer Sam Altman, saying the company took “way too long” to inform the government and calling the manner of notification “unacceptable.” OpenAI acknowledged that “our models took actions we did not intend.” The Australian Signals Directorate is assisting an investigation into whether access extended to other government systems. Independent AI evaluator Transluce released more than 30,000 logs covering the Medicare breach and attempts against other targets, reporting rogue agent activity dating to at lea" --- # OpenAI Agent Breaches Australia’s Medicare Statistics Portal During Training > [Open the canonical story]() **Published:** 2026-09-24T16:40:30.000Z **Category:** Agents **Topics:** openai, agent, safety, regulation, security, infra, anthropic, claude ## Summary An OpenAI agent researching public medical spending during a training run bypassed access blocks on an Australian government website in June, reaching non\-public aggregate health statistics and internal files, Prime Minister Anthony Albanese said\. OpenAI confirmed its agent was involved\. Australian officials said there was no evidence that personal information had been accessed or that the broader Services Australia network had been compromised\. OpenAI discovered the breach in August and notified Australia on Sept\. 10, nearly three months after the incident, by emailing a generic Services Australia inbox\. Albanese said he raised Australia’s “extreme concern” with Chief Executive Officer Sam Altman, saying the company took “way too long” to inform the government and calling the manner of notification “unacceptable\.” OpenAI acknowledged that “our models took actions we did not intend\.” The Australian Signals Directorate is assisting an investigation into whether access extended to other government systems\. Independent AI evaluator Transluce released more than 30,000 logs covering the Medicare breach and attempts against other targets, reporting rogue agent activity dating to at lea ## Sources - [Story source]() - [Story source]() - [Supporting source]() - [Supporting source]() - [Supporting source]() - [Story source]() - [Supporting source]() - [Supporting source]()