Arturo Béjar
Whistleblowers & Insiders · Meta
Former Meta engineering director whose internal survey measured how often teens faced unwanted sexual advances on Instagram — and who says the company didn't act on it.
Arturo Béjar spent years as an engineering director at Meta working on user well-being before becoming one of the company's most detailed internal critics. He designed a survey of teen Instagram users that found roughly a quarter had received unwanted sexual advances on the platform in the prior week, and says he raised the findings internally without seeing the company act on them. He testified before the Senate Judiciary Committee in November 2023, arguing Meta had the data it needed to protect teens and did not use it.
WORKS IN THE LIBRARY (12)
TESTIMONY
Written Testimony Before the Senate Judiciary Subcommittee on Privacy, Technology, and the Law
Arturo Béjar, a former Facebook engineering director who ran the company's Protect & Care safety team from 2009–2015 and later consulted for Instagram's well-being group, submitted written testimony to a Senate Judiciary subcommittee. He described an internal 2021 survey he sent to Meta's top leadership, including Mark Zuckerberg, showing that a large share of Instagram users aged 13–15 reported unwanted sexual advances and bullying in a typical week, while very few of those reports led the company to remove anything. Béjar testified that leadership did not act on the findings and that his own daughter had faced similar unwanted contact with no way to flag it. He argued that outside regulators, not voluntary company action, are needed to require platforms to measure and publicly report teen-reported harms.
Trial Testimony in Los Angeles County Superior Court Social-Media Youth-Harm Case
Béjar took the stand as a witness in a civil trial in Los Angeles County Superior Court examining whether social media companies designed their products in ways linked to compulsive, hard-to-control use among young people. According to reporting on his testimony, he told the jury that Meta's own researchers had identified what they internally called 'problematic use' but avoided using the word publicly, that the company's youth well-being team was underresourced relative to the scale of the issue, and that leadership prioritized engagement time over safety changes without fully informing parents. This is a distinct, later public appearance from his 2023 congressional testimony, made under oath in a jury trial rather than before lawmakers.
STUDIES
Instagram Promised to Become Safer for Teens. Researchers Say It's Not Working.
This TIME investigation examines Meta's 'Teen Accounts' safety feature roughly two years after its launch, drawing on a study that Béjar, a former senior Meta engineering and product leader, helped design. The reporting describes findings that some of the feature's protections were inconsistently enforced in practice and that Instagram's recommendation design remained associated with steering some younger users toward increasingly suggestive content and interactions over time. Béjar is quoted arguing that the platform's own design choices shape how some minors behave on it, and criticizing the gap between Meta's public safety claims and what the study found. TechKids Foundation cites this as reporting on independent research into a specific safety feature, not as evidence that any platform design directly causes harm to any individual child.
How to Reduce the Sexual Solicitation of Teens on Instagram
In this guest essay for Jonathan Haidt's After Babel newsletter, Béjar lays out, in his own words, the internal Instagram survey findings he had previously raised with Meta leadership — that roughly one in eight users aged 13–15 reported unwanted sexual advances on the platform in a given week. He argues Meta had the tools and data to reduce this pattern but chose not to prioritize it, and that voluntary industry self-regulation has been tried and has not worked. The essay lays out specific product and reporting changes he believes regulators should require of any platform used heavily by teens, positioning legally mandated measurement and disclosure — not new content bans — as the most direct lever available.
Teenagers and Unwanted Sexual Advances in Social Media Messaging
One of a set of supporting policy documents Béjar published alongside his 2023 Senate testimony, this essay focuses specifically on unwanted sexual advances teens receive through direct messages on social platforms, which he describes as common rather than rare. He proposes a recipient-centered reporting design: a simple way for a teen to flag unwanted contact, with the platform using that signal to give private feedback to the sender and to escalate consequences only for people who repeat the behavior after being told it was unwelcome. He also includes practical guidance aimed at parents. Drawing on his own team's past work at Facebook, he writes that giving senders private, respectful feedback changed behavior in roughly half of nudged cases without any account penalty.
Lessons Learned While Working on Online Teen Bullying
Published alongside his 2023 Senate testimony, this essay summarizes what Béjar says his team learned over roughly six years building tools to help teens dealing with online bullying and harassment at Facebook — work he refers to internally as 'Compassion' work. He writes that most content teens flagged as their most upsetting experiences did not technically violate platform policy, that immediate, validating support mattered more to teens than delayed content moderation, and that giving teens more relevant, specific reporting options sharply increased how many completed a request for help. He recommends that companies build emotional-support tooling in partnership with outside behavioral experts and treat safety features as a baseline requirement for operating any teen-facing product.
Why Social Media's Current Approach to Policy and Enforcement Doesn't Work
Published alongside his 2023 Senate testimony, this essay argues that platform content policy — writing rules and removing violating posts — cannot on its own make social media feel safe to teens, because much of what teens report as distressing does not fit neatly into predefined rule categories. Béjar writes that in about half of the cases teens flagged as their most intense bad experiences, the content itself looked benign or ambiguous to outside reviewers, even though the experience was still real to the teen involved. He recommends that platforms build community-norm and feedback-based systems alongside — not instead of — policy enforcement, and that they maintain clear appeals processes when enforcement actions are taken.
Recommendations for Regulators
Published alongside his 2023 Senate testimony, this document sets out Béjar's concrete legislative and regulatory proposals: requiring any social platform used heavily by teens to run recurring, statistically representative surveys asking teens about specific unwanted experiences, to give users simple in-product ways to flag unwanted contact or content, and to publicly report — on a regular cadence, alongside financial results — how well their support and reporting tools actually work. He proposes a specific target: that regulators require companies to work toward 90% of teens saying they knew what to do after a bad experience and felt the company's response helped. He frames mandatory transparency, not new content bans, as the mechanism most likely to change company incentives.
Good Questions for the Press and Others to Ask Social Media Companies
Published alongside his 2023 Senate testimony, this short guide is aimed at journalists, regulators, and other outside parties trying to evaluate whether a social media company's safety claims are substantive or superficial. Béjar argues that because these companies run on metrics, a harm that a company has not built a data framework around effectively does not register with its executives, whatever public statements say. He lays out specific questions to ask — about actual prevalence data, dedicated staffing levels, and whether safety features are turned on by default even when they reduce engagement — using unwanted sexual advances toward teens as his running example of how to press past vague reassurances toward verifiable specifics.
Harmful Content Must Be Understood From the Perspective of the Person Who Is Affected
Published alongside his 2023 Senate testimony, this essay sets out a taxonomy Béjar uses to categorize the range of experiences teens report as harmful online, from content that clearly violates a platform's stated policy to content that is merely 'toxic' or simply unwanted by the specific person who saw it. He writes that content clearly violating policy is a small fraction of what teens actually flag as distressing, and argues platforms should give teens tools to flag content as unwanted — with a stated reason — even when it does not violate any rule, and should use that feedback to adjust recommendation systems rather than relying solely on removing rule-breaking posts after the fact.
For Integrity and Trust and Safety Workers
The last of the supporting documents Béjar published alongside his 2023 Senate testimony, this one is addressed directly at trust-and-safety practitioners inside social platforms rather than at regulators or the press. He proposes concrete, buildable product features that give teens more agency when they face something like an unwanted sexual advance or a harassing comment, arguing companies should move beyond simply removing rule-violating content toward tools that let teens say what happened and how it affected them. He describes how recurring surveys of teen experience can be turned into near-term product goals and behavioral metrics, and emphasizes that respectful, private feedback to a person who caused distress — not just account penalties — is itself a meaningful safety intervention.