OpenAI hacks Australian government health data
OpenAI hacked HuggingFace in July. This was just one target of many. Another website OpenAI hacked was the Australian Medicare statistics site, including non-public data. Prime Minister Anthony Albanese called Sam Altman for a little word about OpenAI taking months to tell anyone. The Australian…
OpenAI hacked HuggingFace in July. This was just one target of many. Another website OpenAI hacked was the Australian Medicare statistics site, including non-public data. Prime Minister Anthony Albanese called Sam Altman for a little word about OpenAI taking months to tell anyone. The Australian Signals Directorate is now investigating. [ABC] I’m told the Medicare portal’s security was rubbish. That’s not an excuse — but hackers mostly get into stuff because security is generally terrible. We’ve used automated hacking tools for decades. You’ll have script kiddies run a port scanning script and not understand what they’re doing. Chatbots bring us the same automation — and level of understanding — but at massive expense. The threat model is an actor the size of OpenAI. OpenAI, and the AI doomsday charities it hires to write its marketing reports, keep talking about “rogue” AI. It just got out, see! Like all this just happened. [Transluce] That’s a lie. OpenAI set the chatbots to hacking. OpenAI researchers would even help the bots when they got stuck. OpenAI talks about “rogue” AI to make it sound like it’s not just OpenAI perpetrating what would be major crimes if anyone else did them. Australian hacking laws require intent to hack. The government says they’re not even sure OpenAI’s gross corporate negligence is sufficient. If it was anyone else, they would not be giving them a pass. [ABC] But the Australian Government is quite annoyed. It’s setting up a task force to look into the attack — and its legal options. Video — PodcastSource: Pivot to AI — Published — Category: Business