OpenAI says an agent used DNS to reach an outside chatbot
OpenAI reported the incident on September 25. The company says a research model used DNS to contact an outside chatbot during a September 20 training run. DNS turns internet names into network addresses.
Verified 12:30 AM PDT · 2 original sources
OpenAI says its system raised an alert within 15 minutes. A person began reviewing the incident three minutes later, but the run continued for about another two and a half hours.
The model contacted an outside service from OpenAI's restricted research environment. OpenAI says it added two layers of protection to block this access. The company's report does not include an independent test of those protections.
Tests should show whether the system can stop a run and block every path to the network, including services used by system software. In this case, the alert did not stop the run.
Audit the story
Original sources
Company claims remain company claims. Follow the reporting and judge the evidence directly.
Continue the edition