Auckland dancer Assryia Brady and others had their Instagram accounts suspended over false accusations by Meta. Photo / Supplied
By Lucy Xia of RNZ
A dancer, a dumpling restaurant owner and a makeup artist are among New Zealanders whose Instagram and Facebook accounts have been suspended, as thousands globally call for Meta to be held accountable for what they say are wrongful bans based on false accusations.
More than 37,000 people around the world have signed an online petition advocating for ordinary users impacted – who they said were being silenced by Meta’s “broken AI enforcement systems”.
Meanwhile, Meta said it was using a combination of humans and technology to take down accounts that break its rules – with content such as child sexual abuse material.
In June, it admitted to a glitch in the system that resulted in Facebook groups being wrongly taken down – however, it said that had been fixed.
Auckland dancer Assryia Brady said her Instagram account was suspended in June, after Meta told her she had breached community standards with child exploitation material.
Brady, who mainly posts dance videos, said the claim was baseless and had harmed her reputation.
She said despite being a Meta Verified user, paying $30 a month, she received no support from the company to get her account back.
Freelance makeup artist Tallulah McLean had her Instagram account suspended earlier this month on the grounds of breaching community standards relating to child exploitation, abuse and nudity.
She said the only child photos she had on her account was of her doing makeup for her brother when she was a child.
McLean said the ban was devastating for her livelihood.
“I do all my work as it comes through Instagram, it’s my online portfolio, it’s where I can meet people, like clients, and form connections with other make up brands, PR brands. It’s everything, so this has impacted every form of income for me, like my business, my livelihood, it’s shattered it,” she said.
McLean said her efforts to contact Meta to appeal the ban were met with chatbots and automated responses.
She received two calls from overseas call centres after telling Meta she would be sending legal letters to them. She said the callers told her there was nothing they could do, and she needed to wait.
Meta reinstated McLean’s account within hours after RNZ sent queries about her complaint, but it refused to explain why McLean’s account was suspended.
Auckland dumpling restaurant Sumthin Dumplin also had its business Instagram account suspended for breaching community standards about two months ago despite only posting photos and videos of food and its staff.
Its owner, Shane Liu, said the response from Meta was almost “non-existent” other than chatbot stock responses, even though he was paying about $2000 for his Meta-verified account.
“I didn’t know what to do, what do you do? Did the sales dip? Of course it did. Thank God I built a brand that was beyond just an Instagram page, and it didn’t dip that much, but I was definitely affected.”
Liu said he paid thousands of dollars to a marketing company in Dubai, who eventually helped him reinstate his account.
Meta said it took action on accounts that violated its policies, and people could appeal if they thought it had made a mistake.
“We use a combination of people and technology to find and remove accounts that break our rules,” it said in a statement.
“We’re always working to improve the enforcement of all our policies – including our child nudity and exploitation policy – to help keep our community safe. No system is perfect, which is why we give people the opportunity to appeal decisions if they think we’ve got it wrong,” it added.
The tech giant did not answer RNZ’s questions about how many accounts had been wrongly cut off, and insisted that it had not seen evidence of a significant increase in incorrect enforcement of its rules.
In July, Meta said it was expanding on teen account protections and child safety features, and cracking down on accounts breaking rules.
The BBC reported that Meta was removing 635,000 Instagram and Facebook accounts over sexualised comments and imagery in relation to children.
Artificial intelligence expert from Victoria University Dr Andrew Lensen said the recent shift was surprising, as he thought Meta had in the past taken a conservative approach to suspensions – often not taking down violent and sexual content.
Now they had gone too far in the other direction, he said.
Lensen said while social media platforms had been using automated systems to moderate accounts for a decade, they continued to be increasingly reliant on automated tools.
He said the increasing sophistication of AI-powered moderation tools could be a double-edged sword.
“It’s really hard to increase your detection of the really bad stuff, without also accidentally increasing your mis-detection of accounts that are actually legitimate,” he said.
Lensen said it was hard to know what the motives behind Meta’s recent shift were.
- RNZ