A specialised police unit is using artificial intelligence technology to scour New Zealanders' Facebook, TikTok, Instagram and other social media channels as part of its bolstered intelligence capabilities after the March 15 terrorist attack.
An independent review into "Operation Deans" – the police response to the Christchurch mosque shootings which killed 51 people on March 15, 2019 – made a raft of recommendations into how police could improve future responses.
Tackling the issue of social media, the 49-page review recommended AI-robotics scanning capacity for open source, social media and internet for "key public safety posts", saying it would be "beneficial".
It said such a system was in initial trial stages.
After inquiries from the Herald, New Zealand Police have confirmed that it established an open source intelligence (OSINT) capability following the March 15 attacks.
"The team provides a dedicated capability that assists in the understanding and collection of information from online sources," a police spokeswoman confirmed to the Herald.
The expert unit searches the internet and gathers information "in relation to people, events and issues of interest" to police.
However, it will only be able to access publicly-available information, police say.
The spokeswoman said they will not engage in "any online collection activity that a member of the public could not otherwise lawfully perform themselves with a computer or smartphone".
The tools and systems at the unit's disposal, however, "support the efficient collection of information from the internet".
Security expert Paul Buchanan, a former intelligence analyst for US security agencies, says it took "the wake-up call of March 15" to make police "realise that a lot of unpleasant people are communicating out in the open".
Dr Chris Wilson, programme director of Conflict and Terrorism Studies at University of Auckland, believes the unit will focus on mainstream social media channels like Facebook and YouTube, but also networks favoured by extremists, like Gab, 8kun which was previously known as 8chan, and others.
He believes they will also scour the discussion, chats, and comments of social media pages, blogs, and websites associated with organised groups, as well as invitation-based messaging apps and other forums like Telegram and Riot.
"The team needs to follow and monitor international events, rhetoric, movements, conspiracies overseas – that is what motivates New Zealand extremists. Few extremist ideas are now only national in nature," Wilson said.
The unit's AI algorithms will be tasked with searching for certain keywords, which for white nationalists Wilson thinks might include "Day of the Rope", remigration, and the tree of liberty, while Buchanan thinks they will also look for things like "Knights Templar" references.
"Any time this pops up on a Facebook page, Twitter feed, or Instagram, they're going to have to take a closer look," Buchanan said.
"Their job is actually made a little easier because on the white supremacist side they use certain particular medieval Christian language. And so they've already helped narrow things down, because of the way they talk, much the way jihadists were constantly talking about martyrdom, that sort of thing."
The Christchurch terrorist attacks royal commission's final report, released earlier this month, highlighted the recent move by extremists online.
"One of the most notable changes in the right-wing extremist movement has been its movement from the streets to the internet," the report said.
"In previous decades, the extreme right-wing mostly organised on the streets in gangs or protest movements.
"Today, extremism has substantially, although not completely, moved from physical meetings and street activism to the internet and social media."
The inquiry also noted: "The intelligence function of the New Zealand Police had degraded and from 2015 was not carrying out strategic terrorism threat assessments."
It found that Tarrant, prior to his terrorist attack, occasionally used Facebook to post far right material.
In 2017, he joined The Lads Society's Facebook extremists group Facebook page – and later its private group – where he was an "active contributor", posting far right memes, media articles, and YouTube links, many of which have since been removed from breaching the channel's content agreements.
He also used a gaming site chat room to post numerous links to Reddit posts, Wikipedia pages and YouTube videos that were "far right in nature".