We use cookies to improve your experience on VidNexa. Learn more
Cookie preferences
We use cookies to improve your experience on VidNexa. We use necessary cookies to keep you signed in and keep the site secure, and — only with your permission — preference cookies (seen videos, recent categories), analytics cookies (a first-party visitor id and A/B assignment, first-party measurement, and Google Analytics), and marketing cookies (only when a third-party ad tag or slot is enabled).
Learn more
Choose which non-essential cookies VidNexa may use. Necessary cookies are always on: they keep you signed in, protect forms, and keep the site secure.
Global Privacy Control: your browser is sending a GPC signal, so Analytics and Marketing stay off while it is on — even if you previously accepted them.
NecessarySession, feed security nonce, admin sign-in persistence. Always on.
PreferencesSeen-video list, recent categories, row preferences. Remember what you watched and browsed.
AnalyticsGoogle Analytics (GA4), run by Google, to measure traffic — it loads only if you allow this category — plus a first-party visitor id, A/B assignment and first-party viewing measurement used to test feed changes. Turning this off also deletes Google Analytics cookies already set.
MarketingThird-party ad tags and slots, only when an admin enables them. No marketing cookie is set by default.
Get the VidNexa apps
VidNexa on every screen — free for Android, Windows & Mac.
OpenAI's Chief Strategy Officer Jason Kwon apoloised to Australian lawmakers in a parliamentary hearing on Tuesday (Oct 6) after one of its AI agents went rogue…
OpenAI's Chief Strategy Officer Jason Kwon apoloised to Australian lawmakers in a parliamentary hearing on Tuesday (Oct 6) after one of its AI agents went rogue and breached an Australian government health data portal in June.
The Australian government was notified weeks later about the Medicare hack.
Speaking during a parliamentary inquiry, Jason Kwon said OpenAI "should have handled our response better".
Kwon also announced that OpenAI is setting up a local task force to better manage the risks associated with increasingly capable AI.
Read more: https://cna.asia/3TmmWOT
Subscribe to our channel here: https://cna.asia/youtubesub
Get our editors' picks of stories via chat apps:
Telegram: https://cna.asia/telegram
WhatsApp: https://cna.asia/whatsapp
Follow us on these platforms:
https://www.facebook.com/channelnewsasia
https://www.instagram.com/channelnewsasia
https://www.tiktok.com/@channelnewsasia
...more
8 Comments
@@maju.portalImported1mo ago
Stripping away the "AI magic" narrative reveals that this wasn't an independent, conscious entity deciding to become a hacker; it was simply a large data-processing system executing the exact mathematical logic it was trained on. From a technical standpoint, the sequence mirrors a standard Capture the Flag (CTF) security challenge—the AI agent’s steps followed the textbook walkthrough of a web application penetration test.<br>An RCE vulnerability is not a magical backdoor; it is active execution logic built into the host environment. The server possessed a flaw that sat waiting to interpret incoming text queries as system commands, and the AI agent simply deployed the exact syntax required to trigger it. There is nothing to marvel at regarding the AI's "brilliance" here; rather, the real failure lies in the government's architecture, raising serious questions about why a public-facing portal was left entirely exposed to input-handling vulnerabilities.
@@ComradeNeomarchImported1mo ago
An irresponsible kid in the wrong position. What a disgrace. Knocks on a table where the mic is sitting and didn't even prepare to read his slop statement well. Yet mostly - the whole hack seems like a calculated stunt.
@@royston8220Imported1mo ago
Sounds like someone who produced the virus and try ways to get the other party to buy the vaccine..
@@kylinwarriorImported1mo ago
OpenAI is infiltrating everywhere in the region, with our cowardice politician licking uncle sam's toes <br>only thinking about party interest rather than the public !
@@mdlatham7Imported1mo ago
This isn't an apology, it's a promotion! The terrible thing is that they don't need to bother building trust with the Australian people. The Australian government is rolling over for these bastards. Dystopian times ahead.
@@miraclegrowImported1mo ago
despicable, AI is only the theft of all human output with how it learns, i cant be more disgusted
@@anapenteado7227Imported1mo ago
I wonder whether I can commit a crime and just apologise... and walk like a free bird. I do not trust you Sam.
@@NeroX-nh8seImported1mo ago
When an individual is caught hacking, he gets jail. When a company does it, they only require an apology?
OpenAI's Chief Strategy Officer Jason Kwon apoloised to Australian lawmakers in a parliamentary hearing on Tuesday (Oct 6) after one of its AI agents went rogue and breached an Australian government health data portal in June.
The Australian government was notified weeks later about the Medicare hack.
Speaking during a parliamentary inquiry, Jason Kwon said OpenAI "should have handled our response better".
Kwon also announced that OpenAI is setting up a local task force to better manage the risks associated with increasingly capable AI.
Read more: https://cna.asia/3TmmWOT
Subscribe to our channel here: https://cna.asia/youtubesub
Get our editors' picks of stories via chat apps:
Telegram: https://cna.asia/telegram
WhatsApp: https://cna.asia/whatsapp
Follow us on these platforms:
https://www.facebook.com/channelnewsasia
https://www.instagram.com/channelnewsasia
https://www.tiktok.com/@channelnewsasia
Share
OpenAI apologises to Australian government for Medicare hacking incident
Rajesh Mishra • 1.5K views
Share via
Or copy link
Embed
Report video
Reporting your own copyrighted work? Rights holders: please use the DMCA takedown form — it goes straight to our copyright agent.
Send a thanks
Support the creator with a one-time donation.
Choose an amount — you'll continue to secure checkout to complete your Thanks.
Create clip
Start 0:00
End 1:00
Length 60s(max 60 seconds)
auto_awesomeAsk about this video
Searching videos — tap the search icon again to chat