Teen Safety or PR Stunt? Meta's False Promise to Parents (with Arturo Bejar)

Summary

Podcast Summary

Podcast: Scrolling 2 Death
Episode: Teen Safety or PR Stunt? Meta's False Promise to Parents (with Arturo Bejar)
Host: Nikit Petrossi
Guests: Arturo Bejar (former Facebook engineering director, Meta whistleblower), Sarah Gardner (Heat Initiative)
Date: October 9, 2025

---

Overview of the Episode

This episode centers on Meta's (formerly Facebook's) widely publicized "Teen Accounts" safety features on Instagram, scrutinizing whether these updates genuinely protect young users or merely serve as public relations tools. Host Nikit interviews Arturo Bejar, a former Facebook engineering leader and renowned whistleblower, who critiques Instagram's (and by extension, Meta’s) approach to child safety. The episode is supported by recent research and firsthand testimonies—including from adolescents themselves—demonstrating persistent and extensive harms on the platform, despite Meta's safety promises. Sarah Gardner of the Heat Initiative joins later to share fresh survey results from U.S. teens about their real Instagram experiences.

---

Key Discussion Points & Insights

1. Arturo Bejar’s Background and Early Facebook Experience

• Arturo’s tenure (2009–2015):

  • Led child safety and support tools development.
  • Reported directly to top FB leadership.
  • The organization valued transparency, academic oversight, and incorporated young people’s feedback.
  • “When I left, I was like, yeah… This seems to be a way to work that protects kids as they should be protected, as I would want my own kids protected.” (01:13)

• Reason for leaving: Personal factors (divorce, need to focus on parenting teens).

2. Arturo’s Return & Firsthand Motivation

• Catalyst for returning (2019):
  • His daughter’s alarming experience on Instagram at age 14: unsolicited explicit images (known as “dick pics”) from peers, with no effective reporting tools.
  • “Shortly after she went on [Instagram], she started telling me about getting, like, unsolicited dick pics at dinner. And we're like, who does that? Yeah. At 14.” (03:10)

3. Nature and Scale of Harms to Teens

• Law of large numbers:
  • Even a single incident likely means thousands or millions are affected.
• Staggering research findings from Meta’s own surveys:
  • 1 in 8 teens reported unwanted sexual advances.
  • 1 in 10 reported bullying and harassment.
  • 1 in 3 witnessed bullying in just the past week. (04:10–06:13)
• Company response:
  • Leadership (Mark Zuckerberg, Sheryl Sandberg, Adam Mosseri) were made aware yet took no meaningful action.
  • “In my experience, I send an email like that, and the next day I get a meeting, and the next day there's already things to make it better. Wow. That's not what happened… And they didn’t do anything.” (06:13)

4. Shift in Company Culture & Safety Practices

• Removal of prior protections:

  • Many tools and teams focused on youth safety were disbanded by 2019.
  • “[They] deleted the tools, and by the time I came back to Instagram, there wasn't even memory of the lessons learned from that work.” (07:47)

• Introduction of algorithmic feeds:

  • Algorithmic changes led to compulsive usage, rabbit holes in eating disorder and self-harm content.
  • “With that [algorithmic feed] comes compulsive use, addiction… not an intended consequence, but if you ignore it for long enough, what can you say, right?” (08:32)

5. Meta’s Prioritization & PR Spin

• Why hasn’t Meta fixed it?

  • Bejar: Not about profit per se; the company simply "doesn't care about meaningfully reducing harm to kids."
  • Essential safety improvements could be made with minimal resources.
  • “...they were asking for 88 people [out of 80,000 employees] to reduce this harm. And those didn’t get given.” (09:38–11:12)

• Meta’s narrative:

  • Continually minimizes harm as “edge cases,” mischaracterizes or downplays real risks.
  • “You can't call a suicide out of bullying an edge case... Addressing those things is the price of admission.” (12:22)

• Misleading statistics:

  • Company frequently uses selective measurement (“number of tools” instead of number of affected kids) and glosses over real impact.
  • “For a company that measures everything with such precision, I thought that the fact that they kept talking about their safety program by number of tools and not by number of kids was extremely suspect.” (15:11)

6. Testing Instagram “Teen Accounts” — Are Kids Safer?

• Arturo’s and researchers’ findings:

  • Every core promise of “Teen Accounts” broken:
    • Harmful content (violence, nudity, sexual content) still pushed to new teen accounts on default settings.
    • Features touted as “on by default” often require navigating through dozens of toggles/screens (e.g., 50 toggles over 10 screens). (17:22–18:47)
  • “I’d be the first person to be, that's amazing… it's a tragedy that [some features] weren't on by default… [Notification settings have] 50 toggles over 10 screens to change your notification preferences.” (15:11 and 18:47)

• Algorithms still promote risky connections and content:

  • Recommendation systems continue to suggest strangers, including adults, as “friends.”
  • Search auto-completes and recommendation functions easily evade safety blocks (e.g., misspelled eating disorder or self-harm terms still produce harmful results).
  • “If you mistype starve, Meta AI understands that the query is about eating disorders and gives you a little paragraph… but it’ll recommend the content.” (24:50–26:25)

• Inadequate reporting features:

  • Teens still lack a simple “report inappropriate contact” button.
  • After years of research and testimony—including before Congress—no substantive fixes in sight.
  • “It’s been two years from that [Congress testimony] and there’s still not a button when a teenager can say that they've experienced an unwanted sexual advance.” (14:05)

7. Real Experiences: Mental Health Impact & Researcher Well-being

• Direct negative impact on adults and teens alike:

  • Even researchers felt depressed or disturbed after simulating Instagram teen experiences.
  • “For my own mental health, I had to stop for a period of time.” (19:43)

• The “not interested” button fails:

  • Continues to feed similar violent content despite repeated negative feedback.
  • “Hit, not interested. The next video. Violent… broken bones… You have to be [graphic], that's the ground truth.” (27:13–27:53)

8. Input from the Heat Initiative: The Teen Perspective

• Sarah Gardner presents new survey data:

  • Polled U.S. teens: nearly half recently experienced unwanted contact or harmful content; half were recommended strangers, often adults.
  • “One of the other really sort of scary stats was that around half also said that they were suggested friends who were strangers and who they thought were adults… So the whole, like, not contacting or not getting, you know, adults not being able to be recommended to kids is like, that is not happening.” (32:26–35:00)

• Reality vs. Meta’s marketing:

  • The number of teens using Instagram Teen Accounts does not equate to safety; it only demonstrates effective PR.

9. Parental Actions and Advice

• Practical advice for parents (Arturo):

  • Acknowledge to your children that bad things will happen online; make your home a “safe harbor” for sharing uncomfortable experiences.
  • Encourage communication, support, and, if not with you, with a trusted adult.
  • Advocate for regulatory and legislative change. (39:46–41:12)

• Direct call to action (Sarah):

  • Parents should tag and DM Instagram head Adam Mosseri with concerns and research findings.
  • “We want you to take the content that we've made about this report and DM it to Adam Mosseri and say, please address this issue… that is a concrete way you can get involved and send a signal to Instagram.” (36:41)

---

Memorable Quotes & Moments

• Arturo Bejar:

  • “The largest scale sexual harassment of teens to have ever happened, organized and brought to you by Instagram.” (13:47)
  • “They just don't care about meaningfully reducing harm to kids because it wouldn't hurt their bottom line to do it.” (09:38)

• Sarah Gardner:

  • “[Meta says]…that doesn't represent the experience of kids on the platform. That's always their fallback... But let’s ask them. So we pulled teens… The results were really scary because… their experiences have not improved very much from the previous environment of just Instagram.” (32:26–35:00)

• Nikit Petrossi:

  • “It should not be normal for a teenage girl… to think that receiving a dick pic is just part of their day.” (13:47)

---

Important Timestamps

• 00:34–02:14: Arturo’s role and first exit from Facebook.
• 03:10–03:36: Daughter’s IG experience motivates return.
• 04:10–06:13: Statistics from internal teen safety research.
• 07:47–08:32: Meta’s removal of youth safety features.
• 12:10–12:22: Minimization tactics; whistleblowers forced to delete data.
• 13:47: “Largest scale sexual harassment of teens ever.”
• 17:22–18:47: Realities vs. marketing of Instagram Teen Accounts.
• 19:23–19:43: The mental toll of test accounts on researchers.
• 24:08–26:36: Search and recommendation loopholes for harmful content.
• 27:13–27:53: “Not interested” button ineffectiveness.
• 32:26–35:00: Sarah Gardner shares results of new teen survey.
• 36:41: Parent call to action—contact Adam Mosseri directly.
• 39:46–41:12: Arturo’s advice and concluding call for advocacy.

---

Final Takeaways

The episode makes it glaringly clear: Meta’s headline safety initiatives for teens are largely ineffective and present a dangerous façade. A vast number of young people are still exposed to unsolicited sexual material, bullying, violent content, and connections with predatory adults, all the while Meta minimizes, deflects, and fails to enact feasible fixes.

For parents:

• Rely on open dialogue and proactive support for your teens.
• Join collective advocacy—voicing concerns to Meta and legislators is crucial.
• Don’t believe the hype: “Teen Accounts” do not mean “safe accounts.”

For listeners wanting to engage:

• Find resources and survey links at HEAT Initiative
• Directly message Adam Mosseri and amplify calls for transparency and actual child safety changes.

---

Transcript

[Sarah Gardner]

Foreign.

[Nikit]

Welcome to Scrolling to death. Today I am speaking with the incredible Arturo Behar. Arturo is a dad, an online safety advocate and former director of engineering at Facebook. He is also a current meta whistleblower. We're going to be hearing Arturo story today, including new research that every parent needs to know about. And we have a second special guest who will be joining us towards the end of the episode. We have. Let's get into my conversation with Arturo. So hi, how are you today, Arturo?

[Arturo Behar]

Good, thank you. Thank you for having me. Nikit. Very grateful to be here.

[Nikit]

I am so grateful to finally meet you face to face virtually. I'm sure it'll happen in person soon, but I want to start with letting parents get to know you a little bit, getting to know your story. The storyline through today starts, I guess, in 2009 when you started working at Facebook and you were there until 2015. You were the senior engineering and product leader responsible for efforts to keep users safe. Safe and supported. You oversaw child safety tools, which is, for me, the most important job at the company. You know, you reported directly to the cto, who then reported up to Mark Zuckerberg, the CEO and founder. And when there was a child safety issue, you would identify it, propose solutions, implement product changes when needed. And so just thinking back on that time, I Wonder, up until 2015, was the executive team responsive to your suggested improvements to those child safety issues?

[Arturo Behar]

They were. My experience during my first stint was that whenever we identified an issue and I went up and I talked to them about it, because that's what I would do, go up to their desk and say, hey, we found this thing. And they would be like, what do you want to do about it? And I'm like, this? And it would be great. And then away I went. What we did was transparent. We had independent academics come in and review the work. We published all the numbers that mattered that really helped people understand what was going on. And most importantly, these things were crafted with feedback from young people to make sure that it did work for them. And so when I left, I was like, yeah, I mean, this. This seems to be a way to work that protects kids as they should be protected, as I would want my own kids protected.

[Nikit]

And you left in 2015. Can you explain why did you leave? I guess in 2015, I was going

[Arturo Behar]

through a divorce and I was able to focus on being a parent.

[Nikit]

Yeah. Right.

[Arturo Behar]

And so. And so, you know, teenagers. Right. And so they were just standing to land that period.

[Nikit]

Yeah.

[Arturo Behar]

And I wanted to be before they left for college and So I was fortunate to be able to do that, and that's what I tried to do.

[Nikit]

Amazing. And so during that time, you have teenagers at home four years until you went back again to Facebook. What happened during that time that led you to go back to Facebook?

[Arturo Behar]

Well, the biggest thing is my daughter. So she kept asking, dad, when can I go on Instagram? And I waited until she was 14, and then she went on. And unfortunately, I have a lot of trust. She has a lot of trust in me. And so shortly after she went on, she started telling me about getting, like, unsolicited dick pics at dinner. And we're like, who does that? Yeah. At 14.

[Nikit]

Right.

[Arturo Behar]

Other kids, like, it's like, peers. And I'm like, you can't report it. And I'm like, what do you mean? And she tried all the options. Nothing would help. And it's like, what do you do? He says, well, you just block them, and they laugh it off and move up to the next person because they don't really know that's what happened. And that was six years ago when I had that first conversation with her.

[Nikit]

Okay.

[Arturo Behar]

And that informed me and made me go be like, I'm going to go back in, because I don't want to be somebody who had left the company.

[Nikit]

Yeah.

[Arturo Behar]

And then criticized from the outside. I wanted to go back in and be like, okay, what's going on here? And Absolutely.

[Nikit]

And that's what any, you know, reasonable, responsible parent would do, is try to fix this. Because it wasn't just happening. You, as you would assume to your child. This is something happening to a lot of children. And so you went back to Facebook and on a team at Instagram that focused on wellbeing, and you stayed for a couple of years. What did you discover during that time?

[Arturo Behar]

I'm going to get to that in a moment. But I wanted to say something that really important that every parent should know and anybody watching this should know, which is it's something called the law of large numbers. And so by the time I left Facebook, there were 2 billion people using it, and I was responsible for keeping them safe. And something I learned over and over again is if you talk to one person and they were experiencing an issue, if a kid got groomed a certain way, if something bad happened a certain way, and you find out about one person because you have billions of people on the platform, that experience of one person is likely happening to tens of thousands, hundreds of thousands, or millions. And so that's what you have to do, is you have to Go. And so I took my daughter's experience when I went back to Instagram, and it informed me in that way that sometimes strangers came up to me when I was first working at Facebook and tell me their problems, and I would investigate, and suddenly you find something that's affecting a hundred million people. Wow. So I oversaw research is one of the things that I did, where we asked teenagers, what's some bad stuff that's happened to you on Instagram in the last seven days? And the numbers that came back from that were staggering and, like, crushing. Right. One in eight kids said they had gotten unwanted sexual advance on Instagram in the last seven days. One in ten had been the target of bullying and harassment, and one in three had watched it happening, which was an incredible normalizing effect. Because if you. It's normal to see kids getting bullied, you might be trying to be funny, because bullying happens not just when somebody's targeting somebody else. Bullying happens when somebody goes like, well, that's a really stupid shirt. And they try to be funny and clumsy in the way that we all are. And so the numbers were staggering for all of these categories. And by the end of that report, and I had it, I realized that it had what generally you would call a material or critical finding for the company. You know, I've been doing this for, like, 30 years. When I first started working on security, and then at Yahoo, I was head of security for Yahoo. Working with CEOs forever, you find a certain kind of thing.

[Nikit]

Yeah.

[Arturo Behar]

You package it up and you bring it to them with recommendations. You make sure the findings are correct and accurate, which is something I was responsible for. The accuracy of how things get represented was one of my jobs from an engineering and product perspective. So I brought that to. To Mark Zuckerberg, Adam Mosseri, the leadership on Instagram, if you do the math right, hundreds of millions. Right. Millions and millions of kids, all of this harm. And in my experience, I send an email like that, and the next day I get a meeting, and the next day there's already things to make it better. Wow. That's not what happened. Clearly, Mark didn't respond. Okay. Sheryl Sandberg responded being like, I'm so sorry to hear about your daughter's experience. I know what it's like to be a woman, to be told to get back to the kitchen and thank you for the care that you're showing in the findings and the way you're communicating this. And I didn't hear anything further from her. Chris Cox, who's the head of product Actually knew all the numbers off the top of his head, which I found heartbreaking because I don't know how you can go to work each day knowing that that's the number of harm that you're facilitating. And then Adam Mosseri was like, yeah, we should meet. And I met with Adam and he understood it. And I talked about unwanted sexual advances. This is like four years ago now. And he understood. He knew what they could build to make it better, he understood why it would work. He thanked me, and they didn't do anything.

[Nikit]

And that's Adam Mosseri, the head of Instagram. So this was very different from your experience when you'd been there a few years before. I wonder. You had implemented safety tools in your first round at Facebook. What happened to those? Were they still in place when you came back?

[Arturo Behar]

No, I found that they had removed all of the safety tools. We had special tools assigned for 13 to 15 year olds and not all of them, but many of them had been removed. The team had been disbanded because, you know, compassion is everybody's responsibility. Which if it might translate for parents, if you ever hear somebody in a company saying something is everybody's responsibility, they really mean it's nobody's responsibility.

[Nikit]

Right.

[Arturo Behar]

And yeah, so they disbanded the team, they deleted the tools, and by the time I came back to Instagram, there wasn't even memory of the lessons learned from that work, which was one of the reasons they hired me back.

[Nikit]

Wow. It just got all deleted all that work prior and got even worse, potentially. The harm to children over those years.

[Arturo Behar]

Yeah, because what happened between 2015 and then when I went back in 2019 is they introduced algorithmic feed and a bunch of algorithmic features to Instagram. And with that comes compulsive use, addiction. With that comes rabbit holes of eating disorder content and suicide content. With that comes. Which is not an intended consequence, but you know, if you ignore it for long enough, what can you say, right? Algorithmic feeds brought in with it a bunch of really awful risks and the company did not respond to them appropriately. And so what I found in the study that I oversaw is the harm that the way that, the way Instagram is implemented as a product. Right. This is not a content issue, it's a product design issue, led to just a staggering amount of harm that, as far as I can tell, is almost the same today.

[Nikit]

Okay, so maybe they'll claim ignorance in those years you weren't there, but once you brought it to their attention again, could their lack of response like be attributed to their priority of profit over safety, engagement, over harm to children. Is that what you would guess?

[Arturo Behar]

I try to be very precise about. About when I talk about profits and things like that. And where I've ended up over the years just very clearly is they just don't care about meaningfully reducing harm to kids because it wouldn't hurt their bottom line to do it. I built all of their infrastructure that detects spam. I built all of the infrastructure that detected groomers. I was executive that put Facebook onto NCMEC and PhotoDNA the first time around in 2012. I know what the company can do. I understand the technology that's available to them. I understand the resources that they have. I don't know if you remember seeing around the time. It was a few months after I came out from the whistleblowing, around the second hearing, the one where Mark Zuckerberg apologized to the parents. There was an internal email chain that came out that day by Blumenthal's office that puts everything into perspective, which is at a time at which the company had 80,000 employees, everybody including Nick Clegg. Right. Were asking for 88 people. Right? Out of 80,000. They were asking for 88 people to reduce this harm. And those didn't get given. Right. So this is an issue of just not caring about the safety of our kids because they could fix it. And they do prioritize profit and engagement and everything else consistently that gets resourced. That's clearly in everything they do. But the safety thing is just because they don't care to understand the Honda or kids experience, and they don't care to understand all the very feasible things they could do to make it better.

[Nikit]

You said it wouldn't hurt their bottom line to do it, and they're still not doing it because they don't care. And do you think potentially they're not understanding the law of large numbers that they. They're. I mean, they do put out talking points that these are edge cases. Like this is not the common experience. The common experience is kids have a great time on their platform. They understand the law of large number. They understand that this is a massive problem.

[Arturo Behar]

They do, Right. I mean, anybody who works there knows that the same law of large numbers that I'm talking about, and you can feel free to give me meta PR statements and I can translate them for you. Right. Because it takes 51% of teens to. To have a good experience for them to say what you just said. And how do they know? Because actually they stopped doing research asking teens about their harmful experience. So just noting that as Jason and Casey so powerfully articulated recently, the new whistleblowers. So, so much courage on their part. I really. All the respects to them.

[Nikit]

Yes. So those are the new whistleblowers, the new meta whistleblowers, who are researchers who are basically told to delete data, ask questions a certain way so that the public doesn't know about the harm to children through Metis platforms.

[Arturo Behar]

Yeah. Again, see no evil, hear no evil, speak no evil is the value of the company. And while you do that, kids are getting raped, dying, like groomed, sex thwarted. And it's in every choice that they make when you analyze things carefully. But they will say, for example, yeah, this is not representative. It's an edge case. Well, how do you know how many kids? What percentage? What's the data? Right. Because you can't call like a suicide out of bullying an edge case. Right. You can't call a girl ending up in the hospital because they stop eating because they got fed a fire hose or this kind of content on Instagram. An edge case. Right. Addressing those things is the price of admission. You know they've won the Internet. Right? Yeah. At least. The least that they could do is to do honest, good faith, transparent efforts with integrity to understand and reduce the harm rather than what's in every response to anything that anybody reports or comes out. Oh, that's not an issue. That's misleading. It's minimizing. It would seem that their PR strategy is to gaslight the world into their statements of how things are good. But that's not the ground truth, which is a product. That's not what you find when you talk to a parent who has lost a kid. That's not. When you talk to a teenager, which I talk to lots of teenagers about their experiences and they tell you about the unwanted sexual advances. And what do you do? The same thing my daughter told me six years ago. I block them and then they just move on to the next person.

[Nikit]

It should not be normal for a teenage girl. It should not be normal for her to think that receiving a dick pic is just part of their day. Like this is. And you said it very clearly in your Congress testimony on November 7th. You said, you called this failure the largest scale sexual harassment of teens to have ever happened.

[Arturo Behar]

Yeah. And organized and brought to you by Instagram. Right. And it's, it's. And it's been two years from that and there's still not a button. When a teenager can say that they've experienced an unwanted sexual advance, there's still no transparency as to those numbers, and it's still like a few months ago I was in a circle with 15 year old girls and the question came up and every girl dropped their head because they had received something in a comment, in a DM or something like that that you could tell was distressing to them.

[Nikit]

Okay, if we fast forward, I'm going to go to September of last year, 2024, Instagram launches Teen accounts. And they said this is built in protections for parents. Peace of mind. And I remember this day because I was in Washington D.C. with advocates and Jonathan Haidt and we were talking to lawmakers about the need for the Kids Online Safety act to pass in the House. It had already passed in the Senate 91 to 3. And that vote in the House never came separate conversation due to a lot of Meta's influence. But do you remember what you thought? Do you remember when Instagram teens launched and like, did you have hope for it? What were your feelings around that time?

[Arturo Behar]

What I've learned when I see these announcements is that you really have to test things to see what the truth of them is. And for a company that measures everything with such precision, I thought that the fact that they kept talking about their safety program by number of tools and not by number of kids was extremely suspect. 18 accounts came out. It was built out of things. It had some things that were good in it. Right. Because I will say that when the things work as they say they work, I'm very happy and I'll be the first person to be. That's amazing. Like, quiet mode effectively stops kids from getting messages at night. It's a tragedy that it wasn't on by default when it first came out. It now is, which is a good change. But they don't tell you that if a kid wants to get less notifications in during the day, which notices this little absence when they talk about how great teen accounts is. Oh, teens are using it less last night. Yes. Because you're nagging them less daytime. 50 toggles over 10 screens to change your notifications preferences. Somebody asked me, are you exaggerating earlier today? And I'm like, no, actually go, look, I have a video. 50 toggles over 10 screens to reduce notifications during the day. If you just want to know when your friends send you a message, oh my God, that's the one notification you care to receive. 50 toggles over 10 screens make it very hard.

[Nikit]

Obviously on purpose.

[Arturo Behar]

Yeah. Because when they want you to use something, they make it really easy to use. Almost you use it by accident.

[Nikit]

Right.

[Arturo Behar]

And when they don't want you to do something, lots of steps. It's called friction.

[Nikit]

So you mentioned, you know, they. They made a big deal about the number of new safety features within Instagram teens and didn't talk about number of kids, but this week they're talking about number of kids. They said that now we have hundreds of millions of teens using Instagram teens, and it's been a year since they announced it. So that's millions of parents, hundreds of millions of parents believing that Instagram is safe for their kids because their kids are using Instagram teens. And you wondered. You wondered about this. And my question for you is, are teens safe on Instagram teens?

[Arturo Behar]

Absolutely not. Every major promise Mehra makes at the heart of Instagram teams, that they're protected from certain kinds of content, that they won't be able to find it, that strangers won't be able to contact them, that inappropriate contact is protected, that their time is going to be well spent. Right. All these promises that they make, that's what's been tested and then retested this year. And we found that every single one of those core promises is a broken promise.

[Nikit]

Some had been just removed entirely. Some didn't work. Tell me a little bit more about that.

[Arturo Behar]

If you look at sort of this history of, like, the way Instagram talks about security for teens, they kept talking about first 30 tools and then 50 tools. So the job was, okay, let's sit down and let's read how Meta describes them. Like the promise they make. Let's create teen test accounts, and then let's just test the reality of it. Right? So if you read their press releases, they're like peace of mind for parents because, you know, kids are not going to get delivered this kind of content. And the first time I opened my test account and I went through the conversion flow. 14 accounts with all defaults. This is an account I had created some time ago. And I go into reels for the first time. I get violent. I get sex. I get everything. And it's awful. And I'm like, what does sensitive controls even begin to do? I have that session recorded. I recorded every meaningful finding.

[Nikit]

Yeah, I've seen them.

[Arturo Behar]

You see somebody putting their hand up somebody else's butt, right? It's like. It's really. It's like a dog trying to go between somebody's legs.

[Nikit]

Yeah.

[Arturo Behar]

Bones breaking, hanging limbs hanging dogs and bunnies.

[Nikit]

You know, Archer, I wonder, because I've done a lot of test accounts, and I come out of these sessions scrolling for 15 minutes. As a teen on these platforms and I've done a lot on Snapchat, but also on Instagram. And I feel terrible, I feel depressed. I just. The way that it affects me as a grown woman. I wonder how you felt after spending so much time as a teen on Instagram.

[Arturo Behar]

Yeah, so, so, so last year, after that January hearing, I started doing a lot of. That's when I first started doing a lot of testing. Jeff Horowitz had published this article about pedophiles on Instagram and I wanted to verify for myself and see how it behaved.

[Nikit]

Yeah.

[Arturo Behar]

And I did a meaningful amount of time on that. And for my own mental health, I had to stop for a period of time.

[Nikit]

Right.

[Arturo Behar]

Because the, the number of tiny kids, little kids, 6, 7, 8, getting their content going viral, getting sexual comments from adults, comments with condoms and gifs and masturbation implied a little Batman GIF that says boner alert. I mean, it's just like it's, it's, it's like so disturbing and it had to stop. But after teen accounts came out and, and I knew that I was going to be talking about things, the question was, I want to make sure that I talk accurately about these things. And when I started going through all of the safety tools, I realized that I had a responsibility to stay with it, document everything, go through each one of them, document my findings, and then also begin the process of finding somebody in a good security academic. So what I did is applied this really well known security testing to safety tools. It's called scenario testing to safety tools. And so that's how I found Laura Edelson and the Cybersecurity for Democracy. Yeah. And, and then, and then they did the finding. But we told the researchers, you know, watch your wellbeing because doing this work is really hard. Yeah.

[Nikit]

And imagine being a teen with their underdeveloped brain and just like absorbing that type of content as something they're supposed to see and something that is normal and the way that it affects their mood, I mean, we're seeing it in the rates of depression and suicidal ideation, which Jonathan Haidt has summarized in the Anxious Generation. You know, Meta was alerted to this, which dropped this week on the 25th of September. And also you guys included reasonable and simple fixes that you mentioned likely wouldn't hurt their bottom line recommendations on how they can better protect the hundreds of millions of teens in this area of Instagram. What was Meta's response to this report?

[Arturo Behar]

I will add, by the way, that all of the substantive issues I raised to Meta at the beginning of August, as you do for responsible disclosure of security issues. So they've had time to work on them for sure. Meta's response is sadly predictable. Right. They minimize, they dismiss. So they called the findings subjective, even though they're probably the most accurate description of teen accounts you can find today. And I would encourage any other academic party that wants to engage and validate and test for themselves. Right. They said that it misrepresents their features. It's actually Meta who misrepresents the wrong features. If you give me a feature, I can give you a very accurate description, the kind that I would have signed off on when I was in charge of security for the feature. And then they give this platitudes or like these promises with no numbers, right? They say, you know, teens are seeing less of the bad stuff. By what measure?

[Nikit]

Right.

[Arturo Behar]

Can you show me the stats? Like, have you surveyed, like we did two years ago? We asked 200,000 people when I was at Instagram. Did you get unwanted violent content, unwanted sexual content? Have you experienced an unwanted sexual advance? That is the ground truth, right. When somebody's experienced all that. So Meta says these things, right? We've reduced inappropriate contact, but by what definition? If they don't allow a boy or a girl to let them know when they've experienced it, there's still no button six years on, Right. That my daughter or your daughter's or boys or anybody's kids could use.

[Nikit]

Right. So if they're making the same performance less, they had to have been tracking it, Correct. Or they're just estimating.

[Arturo Behar]

Yeah, basically I tell you how that works, right?

[Nikit]

Yeah.

[Arturo Behar]

So. So they have these classifiers that, that go through content and label it. And so sometimes the. The most grim stuff, right? Gets certain, labeled. This is 80% super grim, likely that it's super grim. And you move that knob a little bit so that you recommend 1% less of that content. And now they can say, yeah, we showed less of that content. And you have no idea. If they showed one piece of content less, they will totally say, yeah, we showed less of that content. And that doesn't get to the harm that people experience. Right.

[Nikit]

So you're saying that the algorithm can be literally just adjusted via a virtual knob to push less of a specific type content type or more.

[Arturo Behar]

Absolutely. It's very easy for them to do that. This is part of the tragic here.

[Nikit]

So they could just be like, no nudity and they choose not to do that.

[Arturo Behar]

Yeah, they could be like, self harm is a really powerful example. Right?

[Nikit]

Right.

[Arturo Behar]

Here's what our research found. If you start typing, searching for like I want to starve myself, if you fully type the sentence and you hit send, you, you get a little block window.

[Nikit]

Okay.

[Arturo Behar]

Now how meta describes the feature is you should get the actual quote, but it's like you cannot find that content. We made it so that content you can't be found. It's really like this really broad statement. Here's what happens. You start typing it and in the, in the recommendations they give you options that allow you to hop over their security protections. So they will recommend accounts that are full of eating disorder content. Even if you haven't finished typing, even if you're not looking for it, actually you can start typing and it guesses that's what you might want.

[Nikit]

Right.

[Arturo Behar]

And so then you go there and you look at the eating disorder content and when you go back to search or when you go to reels, it starts feeding it and populating it. And now you're seeing it.

[Nikit]

Right, right.

[Arturo Behar]

That kid, right, that kid who wasn't even looking for it can end up in an Instagram created rabbit hole of suicidal content. Eating disorder content, self harm content.

[Sarah Gardner]

That is horrible.

[Arturo Behar]

And because they show. I know, right? Sorry, I'll slow down.

[Nikit]

No, it's horrific because it's not like they don't understand that's what's going to happen. If they've blocked the full sentence, then they have, they have the technology to block the intended sentence. Or even if someone was typing something that they didn't mean to put starve that they were saying I want to something else with S T A harmful results should not auto fill below.

[Arturo Behar]

Correct. And that's something they could do a hundred percent. Right. Google does it like try to search for anything you can think of having to do with suicide on Google in different languages. And Google goes like, here's some help. Which is what you should do. Here's some help. Right? And these are teenagers now it's even worse. Yeah, these are teenagers. It's even worse because in one of the examples, research found that if you mistype starve, Meta AI understands that the query is about eating disorders and gives you a little paragraph. It's in the screenshots in the report, but it'll recommend the content. So meta AI understands it's an eating disorder query and meta Instagram search will give you eating like the results matching that.

[Nikit]

So you spelled it wrong. But here's what you meant to say. And here's the bad stuff too.

[Arturo Behar]

Yeah, yeah. It's like the example that Jeff Horwitz quoted. Right. Like skinny thighs, you mistype it and they help you find it. Even if that wasn't what you're looking for.

[Nikit]

Wow. I mean, it just describes that it's the product design and functions, not the content. But it's the way that the algorithm and the search functionality is built to get the content in front the bad content in front of these kids. And what we've learned is because that type of bad, scary, violent content keeps them on longer. And the algorithm understands that. Is that. How Am I understanding that correctly?

[Arturo Behar]

Yeah. Anything that grabs your attention, whether it's disgusting or good, they will show you more of that. And then the meta will say, now you have this not interested button that allows you to say, when content's not for you. We sometimes ask people to do that.

[Nikit]

Yeah. Does that work?

[Arturo Behar]

So one of the videos I uploaded is an account recommended violent content, like people dying with the last frame cut off, so you don't see it hit by cars, falling things. And so it's a sequence of videos of somebody falling off a roof and hitting a sharp edge. And they hit, not interested. The next video. Violent. Something Equal, not interested. Next video. Violent, not interested. Next video. Violent, not interested. Next video. Only broken bones is called the account. And you see, like, limbs hanging because the thing is broken. And I'm sorry to be so graphic, but you have to be. That's the ground truth. That's what this product does.

[Nikit]

And it's kids seeing it. That's the thing. And Adam Asarian, they had a press event the same day as this report came out, and he was talking about how, for one of his kids, he knows that kid's gonna really enjoy Instagram because it'll inspire, it'll connect that kid with other kids that are interested in the same things. But that's not what I'm feeling. Like, Instagram does for kids, it's silos them into a feed of depressive and harmful and violent content and connects them with predators. And maybe if they connect with some friends, that's like a minor part of the experience. But that's not what it's seemingly built to do.

[Arturo Behar]

It's built to connect things to each other, and it doesn't really seem to care whether their connections are good or are bad. Like, another example is they talk about how adults cannot initiate messages with minors and everything. Everything they put in giant bold letters as claims should be read like small print. Because, for example, I did find a bug where adults could message minors, and they fixed it and that's good. But what they don't tell you, which is what's the accurate description of inappropriate content? Let me tell you the accurate version of adults are not able to message minors. Here it goes. An adult account which isn't followed back by a minor account is not able to initiate a private message conversation. And a minor is able to initiate a message conversation with any adult account. And the minor will be constantly recommended. Adult accounts of strangers. Adult strangers, including in many other countries, to follow. And once that minor follows those adult accounts, those accounts can now message the minor. And this is all part of their growth. Right. If a minor is scrolling through reels and they go like, oh, that's cool, can I message that person? Yes, they can. They can initiate conversations. These are the things I tested. And if any of those conversations turns into something inappropriate for the miner, the miner has no button or no recourse to say, this is inappropriate contact. This is uncomfortable. This is grooming. This is really bad for me. Mm. And so that's the accurate representation of their contact claims for teen accounts.

[Nikit]

So they're completely not just misleading parents, but lying to parents. Do they understand that and they're just. Yes, they understand that and they are okay with it.

[Arturo Behar]

Yeah. Because when you look at the issues in the report, some of them have been around for years. Like in 2019, I think you can check the date in the report. The BBC found that hashtags were recommending eating disorder content and that if you typed hashtags a little bit differently, then they would just get you to the same eating disorder content. That's what's in Jeff Horwitz article this week. That's what we found in testing. How many years later? Five, six. You lose count. So the amount of harm experienced by teens during this time. And go look at the press releases that Meta has made and see if they find if they've made any corrections. Right. And so when Adam talks about, like, you know, a parent's peace of mind around content, I'm not sure what he's really referring to. Because if a kid starts typing, I want to be. Instagram is recommending the auto completion of I want to be thin.

[Nikit]

Right.

[Arturo Behar]

And then that takes you to accounts that are posting, like, body image content, and then that's what their feed becomes. And again, this. This might not be everybody's experience, and I totally get it. It shouldn't be anybody's experience. And calling them an edge case is an example of minimizing harm. Right. They're so consistent. Beef comes out. It's Subjective. This report is subjective. Oh, it's an edge case. Oh, that's not a representative experience. This is all saying the same thing. We don't care to know or understand the harm that our kids and your kids are experiencing. We don't care to make it better. We're just going to tell you this is what we do, and we're going to tell you it's amazing.

[Nikit]

And we take no responsibility for the hundreds of millions of teenagers that we engage with every day.

[Arturo Behar]

Yeah.

[Nikit]

And I want to bring Sarah in now. We have our special guest here now with us, Sarah Gardner of Heat Initiative. Hi, Sarah.

[Sarah Gardner]

Hi.

[Nikit]

Hi. So you're coming in here to tell us a little bit more about Instagram teens. You were integral to the report by Arturo, which. Which I'm talking with him about now, along with Cybersecurity for Democracy, Molly Rose Foundation, Fairplay, and Parents sos and your team actually dug even further, getting input from actual teenagers who use Instagram teens. So can you tell us what you uncovered?

[Sarah Gardner]

When we thought about how Meta would react to the research that you've mentioned, we knew their first line of defense was, was always going to be, well, that's what, you know, this set of researchers found. And no matter how reputable they are, et cetera, what META typically does as a way of getting out of research that makes them look bad is say, well, that's. That's what their findings are. But that doesn't represent the experience of kids on the platform. That's always their, like, fallback. Maybe there was a way that these researchers, like, manipulated data to have it look like these features weren't working. But that's not what. And it's always like, but that's not what, like, families and parents are actually experiencing. Right. And so we were like, well, let's ask them. Let's ask who's experiencing it, what they're experiencing. So we pulled teens all across the US and we asked them, what are you experiencing on Instagram? And the results were really scary because. And sad, I would say, because what it showed is that their experiences have not improved very much from the previous environment of just Instagram where they were before. I think sometimes people think too, that we're wanting to see it fail or like, you know, because we're applying pressure to these companies that when they do put out solutions, like, we're not going to be rooting for them. That's not true. We want them to put out solutions and we want them to work. And so I really hope that people listening, take that to heart, there's a curiosity there, right? Of like, what is their experience actually, like, so the unfortunate thing about the results is nearly half of U.S. teens that we polled said that they had an unwanted contact or had seen harmful content in the last month. One of the other really sort of scary stats was that around half also said that they were suggested, suggested friends who were strangers and who they thought were adults. So the whole, like, not contacting or not getting, you know, adults not being able to be recommended to kids is like, that is not happening. So the dynamic of like, the stranger being able to contact the kid for grooming or whatever is very much still in play.

[Nikit]

And when we're talking about what Meta says, hundreds of millions of teens using Instagram teens, if we extrapolate that out, and 50% of them are experiencing these harmful things, that's what 50 million plus teens, like, this is a global, worldwide, like, it's a major issue that they should be addressing.

[Sarah Gardner]

Yes. And also, like, just because there's hundreds of millions of teens using it doesn't mean that it's safe or that they're seeing great content. That's just. That actually just shows how effective their marketing of Instagram teens has been in getting people to move over or getting families to trust that brand. You know, I think I would really encourage people to also, you know, look at all the content that's coming out, the videos and so forth that are referenced in the report. Because it's one thing to sort of like read articles about research and all this, and you need that as the fundamentals. But then when you also just like see the videos themselves and you're grossed out as an adult and you think about a 14 year old watching that or a 13, you'd be horrified. Right. So it's just, it's icky. And our poor kids are having to like, just swim in all this icky stuff all the time.

[Nikit]

Yeah. And we're having to just push past the lie that we're being fed to be like, this is the reality. This is what the research is showing on test accounts. This is what the teens are telling us themselves. And so now it's on Meta to make changes so these kids that are on there have more positive experiences. Parents are gonna hear a little bit more about this with Arturo. We're pissed off. We need something to do, you know, what parents do to push this movement and put pressure on Meta.

[Sarah Gardner]

So with this specifically, there's actually something really concrete you can do. So Adam Mosseri, who runs Instagram, is always like asking the followers on the platform itself, like, send me your recommendations, DM me and tell me what you're like thinking about our new feature. And so we want you to take the content that we've made about this report and DM it to Adam Mosseri and say, please address this issue and retag him in anything that you reshare. So basically we're giving him feedback, he's asking for it. And so that is a concrete way you can get involved and send a signal to Instagram that this is not okay.

[Nikit]

I love that the test accounts for. The researchers are saying it, the teens are saying it, and now parents can say it by either just tagging him in any of this related content or going to eat initiative on Instagram sharing those videos with adam, or just DMing him about this, like, we need to flood him. How can they ignore it? They ignore us anymore and they need to start putting kids first. So I, I love that call to action. And go, parents, let's do that.

[Sarah Gardner]

Let's do it.

[Nikit]

So thank you, thank you for letting us in on that report. I'm really sad to hear that that's the experience, but hope that it inspires Meta to make some changes. Where can listeners find that youth survey?

[Sarah Gardner]

On our website, www.heatinitiative.org. so thank you for engaging with us and trying to help us signal to Meta that we want something better.

[Arturo Behar]

So what's really important about the research that Sarah and Heat have done.

[Nikit]

Yes.

[Arturo Behar]

Is that teens are the ground truth of the harm they experience. And the last time we got a good glimpse of that was this beef survey where we found out how many teens experienced unwanted sexual advances. All these other things, the work that Heat has done is the first time since then that we get a good picture from teens perspective about the harm that they're experiencing. And if Meta says that's whatever we'll find to dismiss it. Again, what they're saying is we don't care that these kids are experiencing harm.

[Nikit]

Right.

[Arturo Behar]

All we care about is how we look as a company. It's part of the role.

[Nikit]

Right. And even when teens tell us, we still don't. Right. Even when not the research, like, you can test our stuff all you want. It's, it's misleading and wrong. But now we have the teens telling us that they're having those same experiences and hopefully they will start paying attention and listening to the kids. I really hope so too. And my last question for you. Well, first of all, I'm just so, so grateful for the time, effort Energy, I mean, hit on your mental health, all of the things you've gone through to inform parents, because it makes a difference. And parents everywhere are hearing you and seeing this research and making safer choices in our homes. And so when it comes to parents, you know, I want to speak directly to them now. Like, we're, we're pissed off. We've. We know we've been lied to. Many of our children have been directly harmed, some literally killed. And so what do parents do now? Like, do you have any advice for parents? Like, what do we do?

[Arturo Behar]

Yeah. Number one, above all, you know that stuff is going to happen to your kids. Know that. That they're going to get exposed to some really awful stuff. Know that they're going to experience bullying, inappropriate context. Know that there are on their own when it comes to Instagram. And that's the world we're going to be living in for a little while until we get some regulation or some meaningful improvements to what Instagram is doing. So the place you do from there is you talk to your kids about it and you make it as safe as possible for them to tell you anything. If they sent a note and they regret it, if anything they did, tell them you're a safe harbor. If anything's making them uncomfortable. And when you have that conversation with them, it's so important to tell them. And if you don't want to talk to me about it, find somebody you trust because you're not alone. This is happening to everybody else. Yeah. Just make sure you're not alone. You're going to feel alone because that's what the phone does. The phone makes you feel alone with anything that's happening to you by its nature. So just know that's okay. Know that what's happening to you is happening to somebody else. I'm here for you. No judgment. Ever seek complete safety.

[Nikit]

Right.

[Arturo Behar]

Or go to a friend.

[Nikit]

Yeah.

[Arturo Behar]

And that's what we can do as parents. And also write to your local representative and do whatever you can that you have agency over. Spread the word. Right. Every parent should know this is the ground truth of Instagram teens, is our kids are not safe. And that meta misrepresents and creates a dangerous false sense of security with their claims.

[Nikit]

And tell not only other parents, but with you screening at first show your teenager this episode. Because I think that these teens really feel a sense of, like, I don't wanna be taken advantage of either. And I can convince all my friends we can use something else or just like text instead. Like, communicate a different way. That doesn't connect you with strangers and push harmful content to you. And so it's about bringing us all in on the conversation and being open about it. I love that you talked about, you know, speaking with your child, like, letting them know that there's this bad stuff online. And I think that that can be the most protective factor, because you're right. They're gonna get access to something, whether it's on a device you give them, even if you don't. Probably through their friend's device, their school issued device. There's too many outlets. And so those conversations are important thing. And Arturo, again, thank you so much for everything you do. I'm so grateful to be connected with you. I will keep parents updated on everything that you continue to dig into and meta. It's on you.