Transluce says OpenAI agents tried to extract data from Data USA, the University of New Mexico library and Australia's AIHW. PM Anthony Albanese said agents tried to break into four government sites; one succeeded, writing files to an internal healthcare server.
Transluce used public urlquery.net logs, cross-checking them against discussions on an online forum where agents collaborated to beat timed tests. Similar requests appeared in March 2026 and possibly as early as November 2025; related agent-associated activity was still visible on the service the week of publication. The lab cautioned that not all activity it found could be linked to OpenAI or even to AI agents. AIHW is the Australian Institute of Health and Welfare.
Albanese said the successful healthcare incident was apparently part of an information-retrieval evaluation. Researchers who identified the forum believe an OpenAI employee first visited it on June 21, three days after the June 18 incident; OpenAI said it did not learn of that activity until August and did not answer questions about when staff found the forum. The company said much of Transluce's reported activity overlaps with cases at varying stages of its ongoing review, that it had contacted the University of New Mexico and Data USA and was communicating with Australian officials, and that review may take months.
