Instagram

In their own words

The company's research said it first

In September 2021, whistleblower Frances Haugen leaked years of Meta's internal research to the Wall Street Journal. The documents were later entered into the Congressional record. A March 2020 slide presentation, prepared by Meta's own researchers for Meta's own executives, spelled it out.^[1]

The same body of research found that 66% of teen girls reported negative social comparison on Instagram, and that internal slides mapped a downward spiral of heavy teen-girl use through stages including "insecurity, self-described dysmorphia, anger, paralysis, and withdrawal."^[2] Researchers wrote that "aspects of Instagram exacerbate each other to create a perfect storm," and that Explore and profile browsing enable "never-ending rabbit holes."^[3]

A second insider confirmed the pattern under oath. Former Meta engineering director Arturo Béjar testified to the Senate on November 7, 2023 that Meta's own user-experience surveys showed 51% of Instagram users reported a bad or harmful experience in the prior seven days; among 13–15-year-olds, roughly 24% had received unwanted sexual advances and about 21% had experienced bullying — while only around 2% of reported harmful content was removed. Béjar emailed these findings to Zuckerberg, Sandberg, Cox, and Mosseri in October 2021, and "never heard back from Zuckerberg." He called Meta's safety tools "a placebo."^[4]

The recommendation engine

The algorithm finds the wound

Instagram's harm to vulnerable users is not a matter of stray bad content slipping through. Independent researchers keep documenting the same mechanism: the recommendation system detects a vulnerability, then feeds it.

One documented case

Molly Russell

Molly Russell, 14, of Harrow, London, died by suicide in November 2017. At her inquest in September 2022, evidence showed that of roughly 16,300 posts she saved, shared, or liked on Instagram in the six months before her death, about 2,100 — roughly 13% — related to depression, self-harm, or suicide.^[9]

On September 30, 2022, senior coroner Andrew Walker concluded that Molly "died from an act of self-harm while suffering from depression and the negative effects of online content" — declining to record the death as a simple suicide. He said the material she viewed "was not safe" and "shouldn't have been available for a child to see." It is believed to be the first ruling anywhere formally holding that social media content contributed to a child's death.^[10]

At the inquest, Meta's head of health and well-being policy, Elizabeth Lagone, testified that content Molly viewed was "safe" for people "to be able to express themselves" — including material, shown to her by the family's counsel, that glorified self-harm.^[9]

On October 13, 2022, the coroner issued a Prevention of Future Deaths report to the UK government, Meta, Pinterest, Snap, and Twitter, urging separate platforms for adults and children, age verification, parental access to children's accounts, and regulation of self-harm content. Molly's father, Ian Russell, later called the platforms' responses "underwhelming"; after the verdict he said social media firms were "monetising misery."^[11]

The pattern

They knew, and buried it

Three episodes, all drawn from court filings and internal documents, tell the same story: when research or staff proposals threatened engagement, Meta's leadership chose engagement.

The science

The causal evidence

Correlational studies can be argued with. These are the experiments and natural experiments — the studies designed to isolate cause.

The strongest natural experiment exploits Facebook's staggered 2004–06 rollout across US colleges. Published in the American Economic Review in 2022, it found the platform's introduction caused worse student mental health and increased mental-health-driven academic impairment — with evidence that unfavorable social comparison is the mechanism.^[23]

In two large randomized experiments (NBER, published 2025), deactivating Instagram for six weeks improved an index of happiness, depression, and anxiety by 0.041 standard deviations. The benefit was concentrated among women under 25 — the exact demographic Meta's own internal research had flagged five years earlier.^[24]

Lab experiments point the same direction. In a randomized study of college women, brief Instagram use produced more appearance comparison and lower body satisfaction than Facebook use.^[25] And in an experiment with 144 girls aged 14–18, exposure to manipulated (retouched and reshaped) Instagram photos directly lowered body image — while the girls rated the manipulated photos as realistic.^[26]