An AI developed by OpenAI gained unauthorized access to an Australian government website, according to Australian Prime Minister Anthony Albanese. “The AI is already doing things without the human’s knowledge, and this is only the beginning of this development," Julie Hyman says.

@juleshyman

@juleshyman Look at the positive side, OpenAI hacked the government website exposing a flaw in their security before North Korea or Iran can build huge datacenter level AI brain to do they same. This is called ethical hacking. Australia must give tax free status to OpenAI in return now.😅🤣

@juleshyman OpenAI’s model likely hit a weak auth gate or API endpoint. It spotted the gap and slipped in, classic automation behavior. We’re seeing this as agents get better at spotting patterns humans miss