As AI surveillance continues to take over, life in Australia could soon look like this.
Your body language being scanned at the airport.
Your behaviour being monitored at a footy match.
The bag you carry through the city watched through a screen by a camera operator.
Right now it may sound far-fetched – but experts say the reality is closer than we think.
A news.com.au investigation is looking into the widespread human impacts that artificial intelligence is already having on our nation in just a few short years.
According to October council papers obtained by news.com.au, the City of Sydney is proposing updating the policy behind their Street Safety Camera Program – which operates more than 100 cameras across the city 24/7.
The draft proposal includes the use of AI “either built into the camera operation or additional add-on software” to alert the council’s security operators to “unusual behaviours”.
The AI-enhanced CCTV will monitor whether an individual leaves or removes an object, whether a vehicle travels in the wrong direction, or abnormal behaviour within a crowd, such as “suddenly gathering or running”.
The proposal also mentions the inclusion of “feature set search tools” such as the identification of sex, age, colour of clothes, hair colour, whether someone is wearing a backpack or bag, and the type and colour of a vehicle.
However, the policy states that “any camera or software analytic technology used will not involve the use of biometric identification systems including facial recognition systems”.
A City of Sydney spokeswoman told news.com.au that the proposal would be placed on public exhibition before any final decisions were made.
“The proposed new draft code of practice would allow the City of Sydney to procure AI-supported software that can help identify public safety risks in CCTV footage,” a City of Sydney spokeswoman said.
“The technology is intended to provide a helpful alert function to the City of Sydney security staff who monitor CCTV in our closed network. A member of that team would still be required to view and verify what is happening before taking action.
“The protection of the rights, freedoms and privacy of our residents, workers and visitors will remain the top priority in any tender for our CCTV program.”
CCTV policies being renewed across Australia
The City of Sydney is not the only council to update its CCTV policy in the age of AI.
In August, the City of Vincent in Perth updated their policy for the first time in 16 years, with the draft proposal including “biometric identification technology”.
It would work to identify “banned individuals from entering or at the point in which they attempt to enter a city facility”.
The council is also exploring the use of AI technology for drowning detection at the local leisure centre as well as the “proactive crime identification” of bikes being stolen by the use of angle grinders.
The policy comes as WA Police continues to trial Live Recognition Technology across the state, scanning faces in Perth CBD, Northbridge and Scarborough this week.
During the first three months of the trial, which began in June and saw 77 operations across 36 locations, more than 900,000 faces were scanned, resulting in 79 arrests and eight incorrect alerts.
Ethics and cybersecurity expert David Tuffley said that while the WA Government had acted in relatively “good faith” during the trial, the fear lied in what “subsequent governments might do” by green lighting the use of this technology.
“We just don’t know how things are going to go down the line,” he said.
“They might decide that a religious gathering of a particular kind might warrant some sort of more surveillance on that.”
Mr Tuffley said surveillance was only going to get more “intense” in future, with AI recognition set to be commonplace at sporting events and airports.
“I see it escalating … but they won’t try to do too much too soon,” he said about facial recognition technology.
“It’s sold to people on the basis of convenience. If you have to queue for 15 minutes or you could just walk straight through … most people would do it and I think that’s how the more intensive surveillance will come.”
He said that keeping human decision making central to these new technologies was vital in keeping them ethical.
“The biggest problem is if the AI is assumed to be capable of acting autonomous without human oversight, then there in lies a lot of trouble,” he said.
False arrests as AI programs expand
The reality is that Australia’s first false, AI-caused arrest is not far away.
In Florida, a man is suing multiple law agencies after a facial recognition program accused him of luring away an underage girl – despite the fact he was almost 500 kilometres away from the incident.
The program detected him as being 93 per cent probability of being the man pictured on CCTV.
It took him almost a year to clear his name.
In December, a Tennessee grandmother spent five months in jail after a facial recognition software incorrectly accused her of fraud crimes in North Dakota.
She had, in fact, never visited the state and was arrested at gunpoint while babysitting her four grandchildren.
Digital forensic expert Stewart Pope told news.com.au that a major concern with AI-enhanced CCTV was that sometimes the technology will “fill in” facial detail that was never there, potentially leading to misidentifications.
“The risk of AI also means that genuine evidence becomes easier to dispute because people go, ‘Oh this is AI-generated’,” he said.
‘Tsunami of crime’
Authorities across Australia are also having to deal with new crime trends ignited by AI.
Acting Australian Federal Police Deputy Commissioner Sandra Booth recently spoke at a federal parliamentary inquiry about a “tsunami” of youth crime due to AI.
She said the radicalisation of young people – which used to take “months or years” – was now taking just days as children continue to veer towards online networks in search of “purpose and belonging”.
The inquiry also heard how potential young offenders were using AI chatbots to research and plan horrific attacks.
Mr Pope said authorities across the world were also seeing a “huge uptick” in fraudulent AI-generated passports and visas, as well as remote mortgage, bank and loan applications.
“For some time, people have been able to make really good photoshopped images of passports … what AI does is it can do a pretty good job of doing that, but it can do that at scale,” he said.
“It might take you hours and hours to make one image as part of a fraudulent visa application … and you’ve got a one per cent success rate.
“With AI, you can very quickly make 100 or 1000 of those, and if your success rate is still one per cent, something is going to work.”
He said security experts also feared the possibility of Australia’s biometric passports being hacked by AI programs.
“We rely on e-passports now with chips in them that have strong encryption, but once quantum computing is widely available … we will get to a world where that encryption can be broken, and our passports are suddenly vulnerable.”
Commander of the NSW Police Cybercrime Squad Matt Craft told news.com.au that it was becoming “very, very hard” to keep updated with the flood of AI-enhanced techniques used by criminals, with state units working together to ensure they are continually upskilling.
“Whatever is coming across, whatever is impacting Australia that needs a nationally consistent approach,” he said.
“Because we can’t always arrest our way out of the problem. As police officers we love nothing more than executing a search warrant, putting handcuffs on, and putting something before the courts
“But when they’re overseas … we need to be more creative.”
Court system overwhelmed
Australia’s court system is also struggling to cope with the onslaught of AI-laden submissions.
In September, one judge was so fed up that she ordered a migration lawyer to pay almost $10,000 in costs for his “unacceptable” use of AI in court proceedings.
Migration lawyer Gurdial Singh Ranji Singh was personally ordered to pay $9600 after 11 authorities mentioned within two submissions he filed to the court were found to be either hallucinated or fictitious.
Federal Court Judge Simone Bingham said the AI errors were only discovered after his submissions were fact-checked by the court.
“It is unacceptable that this occurred,” Judge Bingham said in her decision.
Last year, Judge Judith Gibson, a long-serving judge of the NSW District Court, made a rare declaration about generative AI clogging the courts while handing down her decision in a defamation case.
“This misuse of Gen AI is becoming an increasingly serious problem for the courts, to which the proliferation of recent judgments referring to its use attests,” she said.
“It is a practice that must be stopped.”
Last year, Australia’s top judge, the chief justice of the high court Stephen Gageler, also warned the issue was becoming “unsustainable” in a speech at the Australian Legal Convention in Canberra.
He said that judges were becoming “human filters and human adjudicators” of machine-generated evidence.
“These are existential issues,” Chief Justice Gageler said.
“The need for the Australian judicature to grapple with them is arising as the pace of development of AI is outstripping human capacity to assess and perhaps even to comprehend its potential risks and rewards.”
Get in touch. Sarah.Keoghan@news.com.au