As Australia continues to battle bushfires of unprecedented size and ferocity, a social media disinformation campaign is pushing false conspiracy theories about their cause.
Tweets with the hashtag #ArsonEmergency are coming from a "much higher" proportion of bot-like or troll-like accounts than those with more general bushfire-related hashtags such as #BushfireAustralia or #AustraliaFire, according to initial analysis by Dr Timothy Graham from the Queensland University of Technology (QUT).
Graham came to look at #ArsonEmergency because it was being used by some of the more suspicious-looking individual Twitter accounts he'd been tracking.
"They were really focused in particular on climate denial, and The Greens being responsible for the bushfires, and arson attacks being responsible for the bushfires as well," he told ZDNet on Tuesday.
Those last two are conspiracy theories, he said.
The Greens, or "greenies" in general, have not blocked hazard reduction burning in bushfire-prone areas. In fact they support it.
And while some bushfires are caused by arson, there is no evidence to suggest that the current season has seen a higher level of this crime.
In New South Wales, one of the two states hit hardest by the bushfires so far, only 24 people have been charged with arson. This compares with 53 people charged with failing to comply with a total fire ban, and 47 people who tossed a lit cigarette, or match.
These are just a few examples of the massive amount of bushfire-related misinformation currently circulating online.
Using AI to answer "Bot or not?"
Graham collected a sample of the tweets from January 1 to January 6 using the Python-based scraper twint and ran them through the R package tweetbotornot. That second tool assigns each Twitter account a score from 0 (definitely not a bot) to 1 (definitely a bot).
"Lo and behold, when I ran it through the tweetbotornot fancy AI model, it showed that there's a really high proportion of suspected bots], much higher than we would expect based on... all the other hashtags that are happening at the moment," he said.
Graham warned that the analysis is based on just 315 accounts using the #ArsonEmergency hashtag, compared with 1106 for #BushfireAustralia and 7674 for #AustraliaFire.
He's "at least confident" that it represents a disinformation campaign of some kind. But at this stage he is "less confident" that it's something on the scale of the Russian Internet Research Agency's work during the US elections in 2016.
"I'm not sure whether it's orchestrated, or the extent to which this campaign is being coordinated, but there does appear to be a particular focal point for spreading disinformation about arson in relation to the bushfires," he said.
"I do get a strong sense, based on the evidence so far and based on the analysis I've done, is that it does have all the hallmarks of broader conspiratorial-style thinking [and] far-right populist, extremist discussions online."
More broadly, Graham's major concern is that this whole episode will be "a bit troublesome" for Australia, thanks to what he says is an "unprecedented" amount of disinformation and misinformation.
"A lot of people are confused. There's a lot of anxiety at the moment around people's ability to discern truth, and be able to find out factual information, and to make sense of this tragedy as it unfolds," he said.
"This is a bit of a warning sign for me... I don't think that Australia has seen anything like this before, and it's starting to look a lot like the 60 or 70 examples from other countries around the world where disinformation has become a legitimately and seriously problematic issue for elections and things."
Government warns of fake charities
Meanwhile, as is common during times of disaster, criminals are trying to cash in on people's sympathies.
The Australian government's Scamwatch reports that scammers are pretending to be legitimate well-known charities, creating their own charity names, and impersonating people negatively impacted by the bushfires.
"Scammers are cold-calling, direct messaging and creating fake websites and pages on social media to raise funds," Scamwatch warned on Monday.
"Do not donate via fundraising pages on platforms that do not verify the legitimacy of the fundraiser or that do not guarantee your money will be returned if the page is determined to be fraudulent," they wrote.
"Be careful about crowdfunding requests as these may be fake and also come from scammers."
AI-powered deepfakes, ransomware, IoT, and 5G all mean that protecting your data is about to get a lot harder.
It comes as Labor is fed up with Facebook's lack of action to filter out fake news.
Google has created 3,000 videos using actors and manipulation software to help improve detection.
United Kingdom to have new restrictions imposed within a week due to upcoming general election.