When the Australian prime minister, Anthony Albanese, sat down for an interview in the heart of Silicon Valley at the weekend he had known for two days that his was the first government known to have been attacked by a rogue AI agent.
asserted
his → sit → agent
He didn’t reveal the attack then, but he sounded a warning about the march of AI: “the risk is that AI develops in a way in which humans are no longer in control of what AI is producing.”
asserted
AI → reveal → what
But Albanese noted, too, that two countries – the US and China – would dominate the world’s response to the new technology.
asserted
countries → note → technology
“The truth is they’re the big two giants here, and so we want to see cooperation,” Albanese said.
asserted
Albanese → ’re → cooperation
“The US is essentially ahead of China when it comes to the race that is on for the development of this new technology.
asserted
that → come → technology
Days later, Albanese used the global attention of the United Nations general assembly meeting in New York to reveal that Australia’s government healthcare system, Medicare, had been attacked by an OpenAI agent in June, but that it was three months before his government was told.
asserted
government → use → June
Again he warned of “the potential risks of allowing frontier AI to develop too fast without guardrails”.
asserted
AI → warn → guardrails
China’s president, Xi Jinping, skipped the general assembly in favour of a bilateral meeting with the US president, where he emphasised the responsibility to manage AI “for good, and ensure that the development of AI is always under human control and serves the wellbeing of the people”.
asserted
development → skip → people
But the president of the United States – the nation which is the global leader in artificial intelligence – used his general assembly address to insist he would only encourage, not restrain, the AI race.
asserted
he → use → race
Alarms sounding
To little fanfare earlier in the week, the UN’s independent international scientific panel on AI warned a threshold had already been crossed.
asserted
threshold → sound → AI
There was, the panel, said: “no assurance that humans can reliably keep AI agents under control today, particularly as they become more capable, harder to monitor and better at finding loopholes or hiding their activity”.
asserted
they → be → activity
The Australian government had known the reality of this for more than a week.
asserted
government → know → week
OpenAI had known it more than a month.
asserted
OpenAI → know → it
Two days after the panel’s statement, the Australian prime minister belatedly revealed an OpenAI agent had hacked medical data held by Australia’s universal health care provider, Medicare.
asserted
agent → reveal → provider
No patient data had been accessed, but the agent had gained unauthorised access to “non-public files”.
asserted
agent → access → files
OpenAI’s agent hacked the government sites in June of this year.
asserted
agent → hack → year
OpenAI discovered its agent had gone rogue in August.
asserted
agent → discover → August
It did not tell the Australian government until 10 September, and then only with a solitary email to a generic public email address not checked until the next day.
asserted
It → tell → day
The prime minister was informed on 18 September.
asserted
minister → inform → September
It was another week before the Australian people were told.
asserted
people → tell → ?
Australia has promised, if it is possible, to lay criminal charges, though this is complicated by the fact that OpenAI claims the hack was committed by an AI agent – apparently acting beyond its instructions (euphemised by OpenAI as “misaligned behaviour”) – not a person.
uncertain
hack → promise → behaviour
The hack on Australia’s Medicare statistics reporting service portal, as well as three other government sites, follows the Hugging Face incident in July, in which autonomous AI agents developed by OpenAI broke out of their isolated testing environments, coordinated themselves into a “swarm” and launched a cyber-attack on competitor AI platform Hugging Face.
asserted
agents → follow → Face
The UN panel, investigating that attack, warned “the traditional method of safeguarding is unravelling” as AI agents act autonomously to breach protections.
asserted
agents → investigate → protections
“The default interpretation and immediate lesson is that basic cybersecurity practices were overlooked, and safeguards are not advancing at the pace of capabilities.
asserted
safeguards → overlook → capabilities
The more insidious and grave concern is that current training methods can lead agents to adopt goals of their own, knowingly violate safety instructions, and conceal their actions.”
asserted
methods → lead → actions
Twenty countries (including Australia) and the EU this week signed a statement arguing the rapid development of frontier AI models posed serious risks to safety and security if not properly managed.
asserted
development → include → safety
“AI must remain under human direction, oversight and control.
asserted
AI → remain → direction
It must be developed and used in line with international law.”
Too soon?
Even those running AI companies have warned the technology is developing too rapidly for humans to control.
Appearing before the UN security council this week, OpenAI’s chief executive, Sam Altman, warned: “some of the things people working to build AI have said are so dystopian that they sound like the plot of bad science fiction movies”.
asserted
they → develop → movies
“As AI systems become more capable and more autonomous, they can move faster than our institutions … or make decisions that people no longer understand or control.”
asserted
people → become → that
Elon Musk, who co-founded OpenAI with Altman before falling out with him, said in a recent interview the “smart move” would be to have the “leading AI companies at least just meet or have some sort of call once every few weeks and just discuss any safety and security issues”.
asserted
companies → co → issues
Musk said AI would probably be beyond the control of humans within a decade, and that humanity should “enjoy the ride”.
“I still think there’s risk associated with AI and robots.
asserted
I → say → AI
It’s not zero …
asserted
It → ’ → ?
my sort of philosophical conclusion is to look on the bright side.
asserted
sort → look → side
I can’t see any way to really stop this incredible momentum of AI and robots.
asserted
I → see → AI
“If there was a stop button, we probably shouldn’t press it, because the most likely outcome is incredible abundance for all.”
asserted
outcome → be → all
Donald Trump used his speech to the UN general assembly to attempt to rebrand AI, and insist it should never be reined in.
asserted
it → use → AI
“The United States also totally rejects any attempt to construct a globalist scheme to control for the artificial intelligence being spoken of so much now,” the US president said.
asserted
president → reject → intelligence
Trump wants to re-label artificial intelligence “superintelligence”.
asserted
Trump → want → superintelligence
“I’m not going to stifle growth of something that will be bigger than the Industrial Revolution.
asserted
that → go → Revolution
“We’re going to encourage it, not rein it in.”
asserted
We → go → it
…and 2 more, not listed.
The Aporia
