Former Meta Platforms safety engineer Arturo Bejar, the whistle-blower whose testimony helped build a child-safety case against the tech giant, said its settlement with a coalition of US states has not meaningfully addressed the harms from social media that he described in court.
Bejar told a judge and eight jurors last week that the firm had knowingly under-reported the frequency with which teens were harmed on its platforms, pushed them towards content that made them feel bad about themselves and failed to correctly adjudicate reports of child predation.
On Wednesday, the states announced they and Meta had settled the case. California Attorney General Rob Bonta declared the settlement would “make social media less dangerous for our kids”.
But Bejar said the proposed settlement does not meaningfully address the problems he was asked to testify about.
Meta, on its part, said it is “building on our long-standing efforts to empower parents and support teens”.
“We’re grateful for Mr Bejar’s advocacy and for his testimony in our case,” Bonta said in a statement. “We agree that there is more to be done by legislatures and industry, but believe that the settlement we announced yesterday institutes real and enforceable change, real transparency and real protections for children and teens across the country.”
The settlement involves Meta paying states up to US$18 billion (HK$141 billion) over the next decade. It also includes “remedies” such as limiting how teenagers use Facebook and Instagram.
In some instances, the remedies secured in the case are measures that Meta previously considered, leaked documents show, and rejected after determining they were relatively inconsequential on both user well-being and business grounds.
For example, Meta has agreed to hide likes and reaction counts on posts to prevent young users from comparing their posts’ performance with their peers’.
This effort to combat “negative social comparison” dates back to 2019, when Meta attempted to address its researchers’ concerns that social media might be harming teenagers’ mental health.
The platform tested the effects of hiding like counts in an experiment called “Project Daisy”. It found that the change had at most a modest impact on users’ experience.
“We didn’t observe movements in overall well-being measures,” Facebook employees wrote in a 2020 slide presentation to Mark Zuckerberg. The effects on Meta’s business were similarly slight, with tests indicating that daily user count would likely fall by about 0.09 per cent if the change were adopted.
Meta ultimately allowed users to hide like counts as an opt-in feature.
Concerns about young people’s safety on Meta’s products have been raised for years, with some of the most concerning material emerging from internal Meta research leaked by employees or obtained via legal discovery.
Meta’s researchers had internally flagged algorithms’ tendency to recommend high doses of fitness and beauty content to teens with existing self-esteem issues.
They had also noted the intentional design of features that keep young users scrolling longer than they would independently choose.
Other concerns focused on shortfalls in Meta’s handling of user reports and moderation of inappropriate content. Bejar, for example, oversaw work at Meta that concluded users had bad experiences on Meta’s platforms at rates greater than the company’s official prevalence estimates.
These allegations were central to the states’ case against Meta, but they are addressed only glancingly in the settlement. Instead, the remedies cited by the attorneys general focus on greater parental controls, time limits and boosting efforts to detect minors misrepresenting themselves as adults.
Meta’s ability to identify and remove underage users is itself far from guaranteed as users self-report their ages.
In Australia, where the government banned social media use by teenagers under age 16, regulators have found that eight in 10 young teens remain on social media, prompting authorities to consider additional enforcement.
Reaction from teen mental health experts to the settlement was mixed.
Dr Jane Conron, a clinical psychologist at the Feinberg School of Medicine at Northwestern University in Illinois, said the settlement only requires Meta to offer teens a non-algorithmically curated social media feed on an opt-in basis.
She predicted it would be a feature that would go largely unused.
However, the settlement’s requirement that Meta cap the number of hours teens spend on Instagram and Facebook each day could be meaningful for some users, she said. Some of her young patients are so “emotionally overwhelmed by leaving the app” that they cry when their parents attempt to restrict how much time they spend on it, Conron said.
A built-in usage limit “will help”, she said, adding that she believes the settlement might also change the conversation around social media between teens and their parents.
