Australian federal police say they “had no choice” but to resort to the use of artificial intelligence and are increasingly using this technology to search for seized phones and other devices, given the large amount of data examined in investigations.
The AFP’s director of technology and data strategy, Benjamin Lamont, stated that the investigations carried out by the agency involve an average of 40 terabytes of data. This includes material from the 58,000 referrals it receives a year at its child exploitation centre, while a cyber incident is reported every six minutes.
“So we have no choice but to lean on AI,” he said Wednesday at a Microsoft conference on AI in Sydney.
“It’s beyond the human scale, so we need to start leaning heavily on AI, and we’re using it in a number of areas.”
In addition to being part of the federal government’s trial of Copilot AI assistant technology, the AFP is using Microsoft tools to develop its own custom AI for use within the agency, including the work of translating 6 million emails all in Spanish and examine 7,000 hours of video footage.
“Having… a human being sitting there for 7,000 hours, it’s just not possible. So AI is playing a big role in that,” Lamont said.
One data set the AFP is working on is 10 petabytes (1,024 TB), and a seized individual phone could contain 1 TB of data. Lamont said much of the work the AFP was looking to use AI for was structuring the files obtained to make them easier to process by agents.
“Now when we do a warrant at someone’s house, there are drawers full of old cell phones,” Lamont said. “Now, how do we know that those mobile phones have not been used to commit a crime? “We have to review them and then identify those components and see if there was…any crime there.”
The AFP is also developing artificial intelligence to detect deepfake images and has been trying to figure out how to quarantine, clean and analyze data obtained during investigations, operating in a secure and completely offline environment.
The agency is also exploring whether generative AI could be used to create text summaries of images or videos before officers view them, to prevent them from being unexpectedly exposed to graphic images. The AFP is also studying whether AI could modify such content by converting images to grayscale or removing audio.
The AFP has faced criticism for its use of technology, most notably when its officers used Clearview AI, a facial recognition service created from photographs taken from the internet.
Lamont said the AFP “hasn’t always gotten it right.”
“We’ve had to strengthen our processes internally and I think this… has been really key, because it’s not just set it and forget it,” he said. “As technology evolves and processes evolve… we have to continually look at how we make sure it’s ethical and responsible, and that’s why we’ve created a responsible technology committee within the organization to evaluate emerging technology.”
He said it was also important for the AFP to publicly discuss its use of AI and ensure there was always a human involved in making decisions from the use of AI.