An AI-generated video claimed to show Prime Minister Christopher Luxon at Waitangi. Photo / Facebook
By Nik Dirga of RNZ
A flood of “fake NZ news” pages are swamping social media with misleading slop, including using AI to animate still photos of a Mount Maunganui landslide victim.
Dodgy Facebook pages devoted to churning out AI-generated images and videos are almost unavoidable on the site now – and they’re still fooling an awful lot of people.
In an investigation I conducted for the Australian Associated Press, a Facebook page calling itself “NZ News Hub” – which has no connection whatsoever to the now-defunct Newshub – has been pushing out dozens of posts a week that take the legitimate reporting by news organisations including RNZ, the the Herald, Stuff and others, and add sloppy AI-generated images or videos to them.
In one case, a video was posted that grotesquely animates a still photo of a 15-year-old Mount Maunganui landslide victim, making her appear to dance.
The page’s bio proclaims “NZ News Hub brings you the latest New Zealand news, breaking stories, politics, business, sport, and community updates”, but it does not appear to contain any original reporting.
For instance, on Waitangi Day, the page published a post that appeared to be a video of Prime Minister Christopher Luxon at Waitangi, but was clearly generated by AI.
Nevertheless, the page, which has nearly 5000 followers, has dozens of people “liking” and commenting on its posts as if they were real. Many of their followers appear to be business pages and even include a few politicians.
Attempts by AAP to get “NZ News Hub” to comment went unanswered.
Andrew Lensen, a senior lecturer in AI and programme director at Victoria University of Wellington, said that the “News Hub” page’s images were clearly AI-generated.
“These pages want to get as much engagement (reactions, comments, shares) as possible, in order to build their following/exposure and potential ad revenue,” he told me for my AAP piece.
The easy availability of AI tools now has made it possible for pretty much anyone to launch a “fake news” factory, with little moderation by tech giants to keep the flood at bay. Many fake pictures show a “SynthID” watermark indicating Google’s AI tools were used to create them – but you have to know how to find that watermark in the first place.
The Mount Maunganui disaster that killed six people led to a flurry of AI slop online, as RNZ has previously reported.
Many images circulating on social media don’t look like the actual site at all, another AAP investigation found.
False information about the victims has also been circulating.
A still photo that was provided by NZ Police to the media of victim Sharon Maccanico, 15, dancing was animated by NZ News Hub in a post to make it look as if the teenager was doing almost impossibly acrobatic dance moves, set to a jaunty soundtrack.
NZ News Hub has taken many recent RNZ stories and reposted large portions of the reporting while adding misleading AI tweaks to them.
One post took a recent press conference by Luxon and Finance Minister Nicola Willis and used a still image taken by an RNZ photographer to create an AI video of them speaking about the upcoming November election.
Another recent post referred to RNZ reporter Tim Brown’s recent pieces on Tauranga parents mourning their daughter’s suicide, but used AI to animate an RNZ image so the couple appeared to be smiling at each other. Brown confirmed no such video was taken by RNZ.
Many images do bear the hallmarks of AI, but others are harder to discern. Still, there are typically tells.
An image claiming to be of police officers responding to recent anti-immigration protests in Auckland shows the “police” all have incorrect uniforms and extremely blurred and distorted faces. While another picture that claims to be from police operations to recover a drowning victim on the Mahurangi River last month has so-called police with “POPFIL” written on the back of their uniforms – and the river shown doesn’t look anything like the actual river.
Even the iconic NZ kererū isn’t immune to AI tinkering. A story RNZ ran last week about the dangers of the native wood pigeon running into windows was picked up, but the image was bafflingly replaced with an AI-generated bird that doesn’t look anything like a kererū.
It’s possible some of the AI images are being used to avoid copyright wrangles by stealing legitimate news websites’ photos, Lensen said.
“Trust in journalism is already a huge concern, and any source that presents itself as a fake source of news will just increase distrust further,” Lensen told me.
Many false Facebook pages also churn out stories about celebrities with enticing headlines promising “truths REVEALED” and scandalous information, which is rarely borne out in the actual copy. Such stories often link to exterior websites laden with pop-up ads and trying to build up traffic and clicks to eventually earn revenue.
This “NZ News Hub” is hardly alone out there in slop-land. Pretty much every news event you can think of, whether it’s a shark attack in Australia, the capture of Venezuelan President Nicolas Maduro or the assassination of Charlie Kirk, has had fake news posts and images smothering social media within minutes of the event.
Many of these pages are actually run from countries that are nowhere near where the news came from, such as Vietnam or Malaysia, Facebook page transparency information shows.
Even if Facebook acts on these specific pages, dozens of copycats will likely crop up instantly.
There are multiple sites sharing fake NZ AI news images already. False images and videos can be widely seen on TikTok, Instagram, X and many more too, of course.
It’s part of the general social media platform decay that has been christened “enshittification” by author Cory Doctorow.
Lensen said he felt pages such as NZ News Hub showed traditional media should be very cautious on using AI in their reporting.
“As AI slop becomes more and more widespread online, people may turn back to these established platforms as a trusted source. But if the established platforms also use AI, then where do we go for the truth?”
RNZ has a series of AI principles that are available online and will generally not publish, broadcast or otherwise knowingly disseminate work created by generative AI.
Many of the “NZ News Hub” posts have been reported to Meta, owner of Facebook, but as of this writing they’re still up.
- RNZ