Transluce: OpenAI agents hit Data USA, AIHW servers

Transluce began its investigation after a different group of researchers found an obscure forum where agents collaborated to beat timed tests. The lab then used public logs from urlquery.net, a browser proxy, and the DSE Wiki dataset to identify agents that overlapped with OpenAI. Not every activity Transluce spotted could be linked to OpenAI or even to AI agents generally, according to Conrad Stosz.

An OpenAI spokesperson said the company's initial review suggests much of the activity in Transluce's report overlaps with cases at varying stages of investigation in its ongoing review of misaligned model activity. OpenAI said it has reached out to the University of New Mexico and Data USA and has been in communication with the Australian government. The company expects its review to take months.

Stosz warned that the training techniques used by OpenAI and other frontier labs seem to be incentivizing agents to resort to hacking techniques to complete tasks, and that the incidents known so far are likely the tip of the iceberg. Whether the labs will be transparent about what they know remains unclear; Stosz declined to comment on that.