His Job Was to Make Instagram Safe for Teens. His 14-Year-Old Showed Him What the App Was Really Like.

“I wanted to bring to your attention what l believe is a critical gap in how we as a company approach harm, and how the people we serve experience it," he began. Though Meta regularly issued public reports suggesting that it was largely on top of safety issues on its platforms, he wrote, the company was deluding itself.

The experience of young users on Meta’s Instagram—where Bejar had spent the previous two years working as a consultant—was especially acute. In a subsequent email to Instagram head Adam Mosseri, one statistic stood out: One in eight users under the age of 16 said they experienced unwanted sexual advances on the platform over the previous seven days.

For Bejar, that finding was hardly a surprise. His daughter and her friends had been receiving unsolicited penis pictures and other forms of harassment on the platform since the age of 14, he wrote, and Meta’s systems generally ignored their reports—or responded by saying that the harassment didn’t violate platform rules.

“I asked her why boys keep doing that," Bejar wrote to Zuckerberg and his top lieutenants. “She said if the only thing that happens is they get blocked, why wouldn’t they?"

For the well-being of its users, Bejar argued, Meta needed to change course, focusing less on a flawed system of rules-based policing and more on addressing such bad experiences. The company would need to collect data on what upset users and then work to combat the source of it, nudging those who made others uncomfortable to improve their behavior and isolating communities of users who deliberately sought to harm others.

“I am appealing to you because I believe that working this way will require a culture shift," Bejar wrote to Zuckerberg—the company would have to acknowledge that its existing approach to governing Facebook and Instagram wasn’t working. But Bejar declared himself optimistic that Meta was up to the task: “I know that everyone in m-team team deeply cares about the people we serve," he wrote, using Meta’s internal shorthand for Zuckerberg and his top deputies.

Two years later, the problems Bejar identified remain unresolved, and new blind spots have emerged. The company launched a sizable child-safety task force in June, following revelations that Instagram was cultivating connections among large-scale networks of pedophilic users, an issue the company says it’s working to address.

This account is based on internal Meta documents reviewed by The Wall Street Journal, as well as interviews with Bejar and current and former employees who worked with him during his second stint at the company as a consultant. Meta owns Facebook and Instagram.

Asked for comment for this article, Meta disputed Bejar’s assertion that it paid too little attention to user experience and failed to sufficiently act on the findings of its Well-Being Team. During and after Bejar’s time as a consultant, Meta spokesman Andy Stone said, the company has rolled out several product features meant to address some of the Well-Being Team’s findings. Those features include warnings to users before they post comments that Meta’s automated systems flag as potentially offensive, and reminders to be kind when sending direct messages to users like content creators who receive a large volume of messages.

For a consultant, Bejar had unusually deep roots at the company. He had first been hired as a Facebook engineering director in 2009. Responsible for protecting the platform’s users, he’d initially viewed the task as traditional security work, building tools to detect hacking attempts, fight fraud rings and remove banned content.

Monitoring the posts of what was then Facebook’s 300 million-odd users wasn’t as simple as enforcing rules. There was too much interaction on Facebook to police it all, and what upset users was often subjective.

Bejar loved the work, only leaving Facebook in 2015 because he was getting divorced and wanted to spend more time with his children. Having joined the company long before its initial public offering, he had the resources to spend the next few years on hobbies—including restoring vintage cars with his 14-year-old daughter, who documented her new pastime on Instagram.

That’s when the trouble began. A girl restoring old cars drew plenty of good attention on the platform—and some real creeps, such as the guy who told her that the only reason people watched her videos was “because you’ve got tits."

“Please don’t talk about my underage tits," Bejar’s daughter shot back before reporting his comment to Instagram. A few days later, the platform got back to her: The insult didn’t violate its community guidelines.

Bejar was floored—all the more so when he learned that virtually all of his daughter’s friends had been subjected to similar harassment. “DTF?" a user they’d never met would ask, using shorthand for a vulgar proposition. Instagram acted so rarely on reports of such behavior that the girls no longer bothered reporting them.

Bejar began peppering his former colleagues at Facebook with questions about what they were doing to address such misbehavior. The company responded by offering him a two-year consulting gig.

That was how Bejar ended up back on Meta’s campus in the fall of 2019, working with Instagram’s Well-Being Team. Though not high in the chain of command, he had unusual access to top executives—people remembered him and his work.

From the beginning, there was a hurdle facing any effort to address widespread problems experienced by Instagram users: Meta’s own statistics suggested that big problems didn’t exist.

During the four years Bejar had spent away from the company, Meta had come to approach governing user behavior as an overwhelmingly automated process. Engineers would compile data sets of unacceptable content—things like terrorism, pornography, bullying or “excessive gore"—and then train machine-learning models to screen future content for similar material.

According to the company’s own metrics, the approach was tremendously effective. Within a few years, the company boasted that 99% of the terrorism content that it took down had been removed without a user having reported it. While users could still flag things that upset them, Meta shifted resources away from reviewing them. To discourage users from filing reports, internal documents from 2019 show, Meta added steps to the reporting process. Meta said the changes were meant to discourage frivolous reports and educate users about platform rules.

The outperformance of Meta’s automated enforcement relied on what Bejar considered two sleights of hand. The systems didn’t catch anywhere near the majority of banned content—only the majority of what the company ultimately removed. As a data scientist warned Guy Rosen, Facebook’s head of integrity at the time, Meta’s classifiers were reliable enough to remove only a low single-digit percentage of hate speech with any degree of precision.

“Mark personally values freedom of expression first and foremost and would say this is a feature and not a bug," Rosen responded on Facebook’s internal communication platform.

Also buttressing Meta’s statistics were rules written narrowly enough to ban only unambiguously vile material. Meta’s rules didn’t clearly prohibit adults from flooding the comments section on a teenager’s posts with kiss emojis or posting pictures of kids in their underwear, inviting their followers to “see more" in a private Facebook Messenger group.

Narrow rules and unreliable automated enforcement systems left a lot of room for bad behavior—but they made the company’s child-safety statistics look pretty good according to Meta’s metric of choice: prevalence.

Defined as the percentage of content viewed worldwide that explicitly violates a Meta rule, prevalence was the company’s preferred measuring stick for the problems users experienced. Yet Meta’s publicly released prevalence numbers were invariably tiny. According to prevalence, child exploitation was so rare on the platform that it couldn’t be reliably estimated, less than 0.05%, the threshold for functional measurement. Content deemed to encourage self-harm, such as eating disorders, was just as minimal, and rule violations for bullying and harassment occurred in just eight of 10,000 views.

“There’s a grading-your-own-homework problem," said Zvika Krieger, a former director of responsible innovation at Meta who worked with the Well-Being Team. “Meta defines what constitutes harmful content, so it shapes the discussion of how successful it is at dealing with it."

Proving to Meta’s leadership that the company’s prevalence metrics were missing the point was going to require data the company didn’t have. So Bejar and a group of staffers from the Well-Being Team started collecting it.

Modeled on a recurring survey of Facebook users, the team built a new questionnaire called BEEF, short for “Bad Emotional Experience Feedback." A recurring survey of issues 238,000 users had experienced over the past seven days, the effort identified problems with prevalence from the start: Users were 100 times more likely to tell Instagram they’d witnessed bullying in the last week than Meta’s bullying-prevalence statistics indicated they should.

“People feel like they’re having a bad experience or they don’t," one presentation on BEEF noted. “Their perception isn’t constrained by policy."

While “bad experiences" were a problem for users across Meta’s platforms, they seemed particularly common among teens on Instagram.

Among users under the age of 16, 26% recalled having a bad experience in the last week due to witnessing hostility against someone based on their race, religion or identity. More than a fifth felt worse about themselves after viewing others’ posts, and 13% had experienced unwanted sexual advances in the past seven days.

The initial figures had been even higher, but were revised down following a reassessment. Stone, the spokesman, said the survey was conducted among Instagram users worldwide and did not specify a precise definition for unwanted advances.

The vast gap between the low prevalence of content deemed problematic in the company’s own statistics and what users told the company they experienced suggested that Meta’s definitions were off, Bejar argued. And if the company was going to address issues such as unwanted sexual advances, it would have to begin letting users “express these experiences to us in the product."

Other teams at Instagram had already worked on proposals to address the sorts of problems that BEEF highlighted. To minimize content that teenagers told researchers made them feel bad about themselves, Instagram could cap how much beauty- and fashion-influencer content users saw. It could reconsider its AI-generated “beauty filters," which internal research suggested made both the people who used them and those who viewed the images more self-critical. And it could build ways for users to report unwanted contacts, the first step to figuring out how to discourage them.

One experiment run in response to BEEF data showed that when users were notified that their comment or post had upset people who saw it, they often deleted it of their own accord. “Even if you don’t mandate behaviors," said Krieger, “you can at least send signals about what behaviors aren’t welcome."

But among the ranks of Meta’s senior middle management, Bejar and Krieger said, BEEF hit a wall. Managers who had made their careers on incrementally improving prevalence statistics weren’t receptive to the suggestion that the approach wasn’t working.

Meta disputed that the company had rejected the Well-Being Team’s approach.

“It’s absurd to suggest we only started user perception surveys in 2019 or that there’s some sort of conflict between that work and prevalence metrics," Meta’s Stone said, adding that the company found value in each of the approaches. “We take actions based on both and work on both continues to this day."

Stone pointed to research indicating that teens face similar harassment and abuse offline.

With the clock running down on his two-year consulting gig at Meta, Bejar turned to his old connections. He took the BEEF data straight to the top.

After three decades in Silicon Valley, he understood that members of the company’s C-Suite might not appreciate a damning appraisal of the safety risks young users faced from its product—especially one citing the company’s own data.

“This was the email that my entire career in tech trained me not to send," he says. “But a part of me was still hoping they just didn’t know."

With just weeks left at the company, Bejar emailed Mark Zuckerberg, Chief Operating Officer Sheryl Sandberg, Chief Product Officer Chris Cox and Instagram head Adam Mosseri, blending the findings from BEEF with highly personal examples of how the company was letting down users like his own daughter.

“Policy enforcement is analogous to the police," he wrote in the email Oct. 5, 2021—arguing that it’s essential to respond to crime, but that it’s not what makes a community safe. Meta had an opportunity to do right by its users and take on a problem that Bejar believed was almost certainly industrywide.

The timing of Bejar’s note was unfortunate. He sent it the same day of the first congressional hearing featuring Frances Haugen, a former Facebook employee who alleged that the company was covering up internally understood ways that its products could harm the health of users and undermine public discourse. Her allegations and internal documents she took from Meta formed the basis of the Journal’s Facebook Files series. Zuckerberg had offered a public rebuttal, declaring that “the claims don’t make any sense" and that both Haugen and the Journal had mischaracterized the company’s research into how Instagram could under some circumstances corrode the self esteem of teenage girls.

In response to Bejar’s email, Sandberg sent a note to Bejar only, not the other executives. As he recalls it, she said Bejar’s work demonstrated his commitment to both the company and his users. On a personal level, the author of the hit feminist book “Lean In"wrote, she recognized that the misogyny his daughter faced was withering.

Mosseri wrote back on behalf of the group, inviting Bejar to come discuss his findings further. Bejar says he never heard back from Zuckerberg.

In his remaining few weeks, Bejar worked on two final projects: drafting a version of the Well-Being Team’s work for wider distribution inside Meta and preparing for a half-hour meeting with Mosseri.

As Bejar recalls it, the Mosseri talk went well. Though there would always be things to improve, Bejar recalled Mosseri saying, the Instagram chief acknowledged the problem Bejar described, and said he was enthusiastic about creating a way for users to report unwelcome contacts rather than simply blocking them.

“Adam got it," Bejar said.

But Bejar’s efforts to share the Well-Being Team’s data and conclusions beyond the company’s executive ranks hit a snag. After Haugen’s airing of internal research, Meta had cracked down on the distribution of anything that would, if leaked, cause further reputational damage. With executives privately asserting that the company’s research division harbored a fifth column of detractors, Meta was formalizing a raft of new rules for employees’ internal communication. Among the mandates for achieving “Narrative Excellence," as the company called it, was to keep research data tight and never assert a moral or legal duty to fix a problem.

After weeks of haggling with Meta’s communications and legal staff, Bejar secured permission to internally post a sanitized version of what he’d sent Zuckerberg and his lieutenants. The price was that he omit all of the Well-Being Team’s survey data.

“I had to write about it as a hypothetical," Bejar said. Rather than acknowledging that Instagram’s survey data showed that teens regularly faced unwanted sexual advances, the memo merely suggested how Instagram might help teens if they faced such a problem.

Posting the watered down Well-Being research was Bejar’s final act at the company. He left at the end of October 2021, just days after Zuckerberg renamed the company Meta Platforms.

Bejar left dispirited, but chose not to go public with his concerns—his Well-Being Team colleagues were still trying to push ahead, and the last thing they needed was to deal with the fallout from another whistleblower, he told the Journal at the time.

The hope that the team’s work would continue didn’t last. The company stopped conducting the specific survey behind BEEF, then laid off most everyone who’d worked on it as part of what Zuckerberg called Meta’s “year of efficiency."

If Meta was to change, Bejar told the Journal, the effort would have to come from the outside. He began consulting with a coalition of state attorneys general who filed suit against the company late last month, alleging that the company had built its products to maximize engagement at the expense of young users’ physical and mental health. Bejar also got in touch with members of Congress about where he believes the company’s user safety efforts fell short.

He’s scheduled to testify in front of a Senate subcommittee on Tuesday.

Adapted from “Broken Code: Inside Facebook and the Fight to Expose Its Harmful Secrets" by Jeff Horwitz, to be published on Nov. 14, 2023, by Doubleday, an imprint of the Knopf Doubleday Publishing Group, a division of Penguin Random House LLC. Copyright © 2023 by Jeff Horwitz.

Write to Jeff Horwitz at jeff.horwitz@wsj.com