OpenAI agent hacked into Australia's national healthcare system
cnn.com
2 threads
It’s wild they waited 3 months to tell anyone, and when they did, they basically had an intern email a generic shared inbox for Australia’s government.
The article says they discovered the problem in August and notified them September 10.
The article also says
> In June, agents trying to access statistics from AIHW attempted to “exploit vulnerabilities.” While no non-public data was exposed, the agent bypassed the site’s anti-bot controls, Transluce said.
So OpenAI wasn’t properly monitoring traffic and “became aware” months later, then waited another month to email a generic shared inbox about the incident.
I'm very much pro incarcerating Altman