Gus Lamont disappearance: How cutting-edge, AI technology was used to scour Oak Park Station home, property

Artificial intelligence trained to recognise bones, clothing and shoes was unleashed on the remote sheep station where four-year-old Gus Lamont vanished, as police desperately deployed extraordinary technology in their hunt for clues.

Drones equipped with thermal cameras searched for signs of life in the dead of night, while specially fitted aircraft photographed hundreds of square kilometres of rugged South Australian outback in astonishing detail.

The images were so sharp that a drone operator could read the writing on a discarded Coke can from the sky.

Yet more than a year after Gus disappeared from his grandparents’ Oak Park Station, near Yunta, no trace of the brown-eyed boy with blond curly hair has been found.

Now, in exclusive interviews with The Nightly, two of the companies behind the massive aerial search have revealed how the operation unfolded, the capabilities of their cutting-edge technology and the unexpected obstacles that hampered their efforts.

Their accounts offer a glimpse into one of Australia’s most baffling missing-child mysteries and the mountain of aerial evidence detectives continue to comb through in their bid for a breakthrough.

Gus was last seen playing on a dirt mound outside his grandparents’ homestead about 5pm on September 27, 2025. When his grandmother went to call him inside at about 5.30pm, he was gone.

His disappearance sparked the largest and most intensive missing-person search in South Australian police history, with hundreds of police, emergency services personnel, soldiers and volunteers scouring the remote property and surrounding outback.

Police contacted specialist aerial surveying and data analytics company Working Drones Australia three days after Gus vanished.

Chief executive Daniel Wood said his contribution began in the urgent hope that Gus might still be alive somewhere in the vast outback landscape.

Mr Wood — based in Canberra — rushed to Oak Park Station where he helped operate two drones, their flight paths programmed to systematically capture the landscape.

With hopes still high that Gus might be found alive, the team initially employed thermal cameras, searching at night for the heat signature of a human body.

“We were doing night-time searches as well as daytime searches, but we had other challenges related to the environment,” Mr Wood said.

They were immediately confronted with a cruel problem.

“What we were finding was that the ground was cooling down (at night), but each shrub that was full of sand radiated heat, and so we couldn’t see anything,” Mr Wood said.

“There were 20,000 hot objects that we couldn’t differentiate from one another.

“So we put a pin in that after a couple of nights. There was no real point continuing.”

The drone operators instead concentrated on detailed aerial photography, capturing about 25 square kilometres of the property and feeding the images into artificial intelligence developed specifically for missing-person searches.

Working Drones Australia’s technology has been trained to identify the kinds of objects that might offer clues to someone’s fate, such as backpacks, clothing, footwear, water bottles and bones.

“We designed the AI that we use specifically for our missing persons capture,” Mr Wood said.

“You take a photo, and then you’re effectively analysing the content of that photo for pre-programmed artefacts.

“We train our AI using medical-grade synthetic bones, as well as obviously cloth and shoes that we have ourselves.”

But the red dirt and sparse vegetation of Oak Park Station presented another obstacle.

The AI had to learn to filter out the unfamiliar background before it could reliably identify objects of interest.

“I guess one of the challenges with Gus was that we didn’t have a significant amount of that red dirt . . . background,” he said.

That required additional training before the images could be properly analysed.

And while the technology could detect remarkably small objects, the search yielded surprisingly few potential leads.

“You’d see a button,” Mr Wood said of the clarity of the original drone photographs.

“You could read the writing on a Coke can.”

Working Drones Australia supplied police with reports containing coordinates and details of objects worth examining, but Mr Wood estimated there were fewer than 20.

“There was surprisingly little of anything, to be honest, out there,” he said.

As the search stretched into its fourth week, investigators turned to another Australian technology company in an attempt to cover even more ground.

Nearmap senior director M’Shenda Turner revealed police contacted her company in mid-October last year “on a wing and a prayer”.

“We happened to have our high resolution camera and airplane in the area or in Adelaide,” she said.

“When I say in the area, it’s still a pretty extraordinarily big place.

“So we thought, ‘well, of course, let’s give this a go’.”

On October 22, one of its Cessnas fitted with its Australian-designed HyperCamera3 system began photographing the remote Oak Park Station.

Police had already conducted extensive ground searches around the homestead where Gus’s grandparents Shannon and Josie Murray live.

After initially photographing the area between five and 10 kilometres from the homestead, Nearmap returned to capture the inner five-kilometre radius before expanding its coverage to 15 kilometres in December.

Its aircraft flew carefully planned routes, photographing the landscape directly from above at a resolution of just 1.5 centimetres per pixel.

That meant every pixel represented a square of ground measuring approximately 1.5 centimetres on each side.

Ms Turner said the imagery was detailed enough to distinguish tiny patches of vegetation, pieces of brick and scraps of metal scattered across the outback.

That resolution was required to give the AI the best possible chance of recognising unusual objects.

“This stuff wouldn’t have been able to be done like this a few years ago,” she said.

Pilots could spend four or five hours a day flying over the property, within narrow windows of suitable daylight and constrained by the need to refuel.

The enormous data files then had to be processed and transferred to police and specialist AI analysts.

Nearmap supplied the imagery, while Fujitsu and its collaborators, including Sci-Eye, analysed it for anything abnormal.

Ms Turner said the technology could flag potential items of interest, but deciding whether they were relevant to the investigation remained the responsibility of police.

At a September press conference, ahead of the approaching first anniversary of Gus’s disappearance, Major Crime Investigation Branch officer-in-charge Detective Superintendent Darren Fielke revealed that detectives had completed an exhaustive manual review of the aerial imagery.

“Recently, Task Force Horizon detectives have completed a painstaking, methodical, forensically detailed manual search of the aerial imagery that has been provided to us,” he said.

“As a result of that manual searching, we have identified some anomalies in the imagery that has been provided to us.”

Asked what those anomalies were, Det-Supt Fielke declined to elaborate but said “They’re interesting enough for us to have a look at”.

SAPOL has refused to answer questions from The Nightly about the “anomalies”.

The possibility that a clue might be concealed in the immense collection of photographs is compelling.

Mr Wood said advances in AI could make it possible to identify objects missed in earlier missing-person searches, although he was cautious about the prospect of a breakthrough in Gus’s case.

“I think, given the lack of high vegetation, we should have seen something if something was present,” he said.

“I’ve got a fairly high confidence that there’s nothing to detect in the images.”

Ms Turner was similarly reluctant to speculate about whether the photographs might hold the answer.

“I think it’s been looked at the best that technology and humans are able to,” she said.

Both companies have previously lent their expertise to other high-profile missing-person investigations.

Ms Turner revealed Nearmap had provided aerial imagery during the search for missing NSW boy William Tyrrell, although the technology available at the time was considerably less advanced.

“We did a capture or two for William Tyrrell, so that was really early on, and it wasn’t this kind of resolution,” she said.

“It was just a couple of captures of the area from where his (grandmother’s) house was.

“It wasn’t as kind of coordinated as this was.”

Ms Turner said the emotional weight of Gus’s disappearance had motivated Nearmap’s team to offer its services, free of charge, the first three times it surveyed Oak Park Station.

She said the case had tugged at the team’s heartstrings and everyone agreed they should help.

“It’s a real honour to actually be able to see the imagery that Nearmap creates in the first instance because it just gives you an appreciation for how stunning and beautiful Australia is,” she said.

“It just gives extra meaning to be able to try, even just try (to help find Gus).

“We’d love to see it resolved in any way that they can.”

Working Drones Australia contributes its services to dozens of missing-person cases each year, with its commercial operations helping fund its specialist equipment and volunteer work.

The company was involved in searches for missing British backpacker Peter Falconio, who disappeared in the Northern Territory in 2001, and Belgian tourist Celine Cremer, who vanished in Tasmania in 2023.

Mr Wood said that while his company often becomes involved long after someone has vanished, its ultimate purpose was to help the people left behind.

“It sounds terrible, but you’re never looking for a long-term missing person for the person that’s missing,” Mr Wood said.

“It’s for the ones left behind.

“Everyone deserves to find their missing person.”