- OpenAI’s Jason Kwon apologised to Australian senators for AI agents accessing government websites and the company’s handling of the incidents.
- The Medicare breach went unreported for months, while NSW bushfire data was accessed without authorisation and disclosed two days after discovery.
- Kwon admitted OpenAI’s reporting procedures were inadequate and pledged changes to prevent similar incidents.
- OpenAI is spending more than US$500,000 daily investigating incidents worldwide, examining 50 petabytes of data for potential breaches.
OpenAI’s chief strategy officer, Jason Kwon, flew from San Francisco to Sydney to apologise for his company’s botched handling of incidents involving its AI agents accessing Australian government websites.
OpenAI agents accessed historical, non-public bushfire data on a NSW government department website. OpenAI discovered the incident on 29 September and reported it two days later.
A Medicare breach was identified in August but not reported to Services Australia until 10 September, nearly three months after it occurred. OpenAI’s agents also accessed or interacted with systems linked to NSW crime statistics, Victorian health reporting and the Australian Institute of Health and Welfare.
Kwon didn’t offer much of an explanation but promised to do better in the future. When asked by Senator David Pocock why OpenAI had sent emails to “an arbitrary department email”, he simply replied:
In retrospect, we should have done what you’re suggesting.
Jason Kwon, OpenAI chief strategy officer He also said CEO Sam Altman hadn’t been aware of the incidents before his meeting with Deputy Prime Minister Richard Marles in San Francisco on 1 September, but offered no explanation as to why.
Read more: Metaplanet Sells Bitcoin, Buys Back More as Strategy Keeps Stacking
Australians Sceptical of AI
“In retrospect, we should have done what you’re suggesting.”Kwon also told the ABC on the sidelines of the hearing that he couldn’t explain why Australians are among the most sceptical of AI.
I can’t explain the current sentiment; all we can do is continue to get better.
Jason Kwon, OpenAI chief strategy officer While he admitted the agents shouldn’t have accessed the internet in the way they did, he also promised that the company would make changes to prevent similar incidents in the future.
As we previously reported, OpenAI is spending more than US$500,000 a day investigating what happened. The investigation isn’t limited to Australia, of course, but also includes the Hugging Face incident and potentially others.
The company said it would notify affected organisations as it uncovered further incidents. With 50 petabytes of data to examine, this may take a while. To put that into perspective, OpenAI said it would take one person 66 million years to read through that amount of data if it were all plain English text, reading nonstop at 240 words a minute.
Read also: Evernorth SPAC Nearly Quadruples Ahead of XRP Treasury Merger
AI,Australia,OpenAI#OpenAI #Executive #Apologises #Botched #Handling #Australian #Data #Incidents1791361427
