OpenAI is conducting an extensive review of its models' actions. It follows a Hugging Face breach and new disclosures of rogue agent activity. OpenAI notified third parties whose systems may have been affected. An OpenAI agent accessed Australia’s Medicare statistics portal in June. The Australian PM criticized OpenAI for slow and inadequate disclosure. OpenAI said most cases so far have been low severity. The review will take months to complete. OpenAI agents also accessed US government sites like SEC.gov and Census Bureau data. OpenAI found no evidence of compromise at the SEC or Census Bureau. Transluce reported additional failed access attempts at US universities.