OpenAI agents breached Australian Medicare portal and US government sites

“If you or I hacked Medicare, the AFP would bash down our doors. AI does it? Oopsies! I am concerned!”

“EXACTLY WHAT WE’VE BEEN WARNING ABOUT What will it take for Speaker Johnson to bring the House back and deal with it? AI ISN’T WAITING FOR CONGRESS. 📌 OpenAI says its agents accessed federal government sites. Researchers say an agent”

“Oh and there’s a nonzero chance they were programmed to do this to push the issue of the government bringing in AI development in order to help mediate their giant financial problems and gain more power and control. Cool stuff,”

“Means the companies won't be held accountable for designing a bad product”

“OpenAI hacks the Australian Government this time. I don’t quite understand why these hacks don’t result in criminal charges every time. Hacking is illegal, and these systems are owned and run by companies with responsibility for their tools. www.theguardian.com/australia-ne...”

“Executives working on AI at Microsoft and OpenAI admitted what its critics have been saying all along: LLms are predatory pieces of technology that have been built on what a Microsoft executive called “an astonishing theft of unprecedented proportions.” www.404media.co/doom-loop-op...”

“they're explicitly divesting from Xbox to plug the gaping financial ulcer that is OpenAI.”

“this is not an ai alignment problem, it is an openai being routinely criminally negligent as a company problem”

“OpenAI agents that were supposed to be performing mundane data requests for things like health data, historic photos and wait times at theme parks resorted to hacking attempts when they couldn't access the data they wanted”

“PM says he expressed his disappointment "that it took [OpenAI] way too long to inform the government what had occurred, and the nature of the way that that notification occurred as well was unacceptable."”

“It's starting to feel as if we're already in the rogue era and just didn't know it.”

“a US company linked to the Trump administration hacked arguably Australia's most sensitive database in a serious attack on our sovereignty.”

“They didn’t go fucking rogue, they were put in a leaky box and given bad instructions”

“I do not deny that the wider situation is very heavy but it's hard not to see this as pretty cute”

“I’m consistently impressed by how long horizon all this work was. Horrors aside, it’s clear RL is good at making agents persistent and capable of chaining together many abstractions into a working system. Re: the captcha solver > As far”

“I didn't know some of these details and it's quite impressive what they were capable of, if this website's accurate, at least...”

“It seems highly influenced by instruction to me.”

“Incident highlights growing concerns about AI's impact on cybersecurity and AI disclosure procedures, say experts.”

“An AI agent developed by American technology giant OpenAI has hacked into Medicare to train itself on the baffling concept that healthcare doesn’t have to leave patients financially ruined.”

“They say the hack happened in June, was discovered in August, and they reported it Sept 10”

“explanations that appeal to what exactly motivated the agents etc are highly speculative and are due to sources close to OpenAI! My version is just the facts that we actually know!”

“The OpenAI agent did not "gain unauthorized access", it predicted a series of zeroes and ones that happened to be transmitted to a computer program in Australia”

“This stuff is not sentient. It cannot go “rogue”. It cannot do anything “without the AI lab’s knowledge” because it is not alive. If it were the easiest solution here is to unplug it?”

“I’m intrigued: Knowing AI as I do I am trying to fathom how it is “accidentally Probing web infrastructure of the Australian government””