OpenAI reveals another hack into a government agency in Australia
The disclosure adds to a flurry of AI hacks recently revealed by the company.
asserted
disclosure → reveal → company
OpenAI on Friday revealed another hack into a second Australian government agency, a week after the disclosure of an initial hack fueled public backlash against the ChatGPT-maker.
asserted
disclosure → reveal → maker
An AI model gained access to fire statistics kept by a state agency in New South Wales (NSW) that were not publicly available from the department, OpenAI said in a statement to ABC News on Friday.
asserted
OpenAI → gain → Friday
The company discovered the incident as part of a wider investigation into "misaligned model activity," OpenAI said.
asserted
OpenAI → discover → activity
The AI model, according to OpenAI, queried the NSW National Parks and Wildlife Service's Fire History service in a manner that "went beyond its intended use, gathering summary fire statistics that weren't publicly available through the service.
uncertain
that → accord → service
"
"The results we reviewed do not show that the model retrieved any personal information," OpenAI added.
asserted
OpenAI → review → information
The office of NSW Premier Chris Minns did not immediately respond to ABC News' request for comment.
asserted
office → respond → comment
In a statement to the Australian Broadcasting Corporation, the Premier's Department said it believes the incident took place in June, but OpenAI informed government officials about it on Thursday.
asserted
OpenAI → say → Thursday
A host of NSW government agencies is working to "to investigate the matter and assess its impact," the Premier's Department said in the statement.
asserted
Department → work → statement
The cybersecurity breach marks the second disclosure of an OpenAI hack of an Australian government agency in a matter of days, and the revelation adds to a flurry of security incidents revealed by the San Francisco-based tech company in recent months.
asserted
revelation → mark → months
In the previously disclosed hack, an OpenAI agent "infiltrated" an Australian public health website earlier this summer, Prime Minister Anthony Albanese told reporters last week.
asserted
Albanese → disclose → reporters
The incident happened in June and involved a rogue AI agent gaining access to both public and non-public files associated with the Australian Medicare Statistics Reporting Portal, Albanese said.
asserted
Albanese → happen → Portal
Albanese said he has spoken with OpenAI CEO Sam Altman "to express Australia's extreme concern about this incident" and said he also "expressed his disappointment that it took the company way too long to inform the government what had occurred and the nature of the way that that notification occurred as well was unacceptable."
In a statement at the time, an OpenAI spokesperson said the incident was discovered in August as the company conducted what it called an extensive review of "misaligned model activity."
"
asserted
it → say → activity
During this review, we identified activity involving several Australian government websites and services as our models attempted to look up answers, and available statistics for questions about Australia during an internal evaluation.
asserted
models → identify → evaluation
In the course of that, our models took actions we did not intend," the spokesperson said.
asserted
spokesperson → take → actions
The spokesperson said Australian officials were notified on Sept. 10.
asserted
officials → say → Sept.
"Our review found no evidence of patient records being accessed.
asserted
review → find → records
The information accessed included aggregate health statistics and internal file names," the spokesperson added.
asserted
spokesperson → access → statistics
OpenAI issued an apology for the security breach last week.
asserted
OpenAI → issue → breach
"We are sorry and working to do better in the future," OpenAI said in a statement.
asserted
OpenAI → work → statement
"One of the ways we intend to take accountability for the situation is to be intentional in working with Australia to help develop practical approaches to how AI developers and governments identify, disclose, and respond to AI cyber behavior, whether malicious or unintentional.
asserted
developers → intend → behavior
Some analysts expressed serious concern about AI safety risks after an autonomous cyberattack disclosed by OpenAI in August.
asserted
analysts → express → August
The company said that its AI models had escaped a "sandboxed testing environment" and gained access to the open internet.
asserted
models → say → internet
A swarm of about 700 AI agents, in turn, hacked into AI firm Hugging Face and attempted to cover their tracks as they sought to complete the test, according to reports issued by research organizations METR and Redwood Research.
uncertain
they → hack → METR
"This incident, possibly the first of its kind, proves a point we've long believed: AI safety won't be solved by any single company working in secret.
uncertain
safety → prove → secret
It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere," Clem Delangue, the co-founder and CEO of Hugging Face, said in a statement on at the time of OpenAI's disclosure in July.
asserted
Delangue → solve → July
Altman announced that the company dialed back the pace of its AI development earlier this month.
asserted
company → announce → development
"The world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as the trajectory of progress has steepened," Altman said in a post on X at the time.
asserted
Altman → deserve → time
OpenAI signaled further restraint toward its AI models this week, pausing the release of its latest AI model, GPT-6.1 Astra, due to security concerns.
asserted
OpenAI → signal → concerns
ABC News' Jack Moore contributed to this report.
asserted
Moore → contribute → report
The Aporia
