Valuetainment - November 13, 2026


Facebook Whistleblower Blasts Refusal to Fund 45 More Child Safety Staff

Facebook Whistleblower Explains How Algorithms Can Fuel Civil Wars

Episode Stats


Length

17 minutes

Words per minute

208.41

Word count

3,701

Sentence count

154

Harmful content

Misogyny

3

sentences flagged

Toxicity

7

sentences flagged

Hate speech

3

sentences flagged


Transcript

Transcript generated with Whisper (turbo).
Misogyny classifications generated with MilaNLProc/bert-base-uncased-ear-misogyny .
Toxicity classifications generated with s-nlp/roberta_toxicity_classifier .
Hate speech classifications generated with facebook/roberta-hate-speech-dynabench-r4-target .
00:00:00 maybe walk me through when you were having a meeting with wall street journal and you're like
00:00:04 ah should i do this should i not i don't want to do it i know what you're asking me maybe you know
00:00:08 i should and then all of a sudden you're like i have all the documents i took all the pictures
00:00:12 i'm coming out i think the world needs to know what was that process sure so when i originally
00:00:17 came to jeff my main issue was i we we had done internal research where we had talked to political
00:00:22 parties in europe and they had said hey we are intentionally running different political positions
00:00:27 post you changing the algorithm.
00:00:29 So the idea that a private company can make a decision
00:00:32 about what information we get to see or not see
00:00:34 and that public entities, public political parties
00:00:37 are saying, we don't give voters the same options
00:00:40 because of choices you made in Menlo Park.
00:00:42 That seems like a thing that the public had a right to know.
00:00:45 So initially, that was my big focus,
00:00:47 was saying like, hey, they know what's going on.
00:00:49 They're not telling people our elections are being influenced
00:00:51 even in the options that get to choose.
00:00:55 I worked with Jeff.
00:00:57 We always try to leave this part slightly ambiguous.
00:00:59 I've been told by my lawyers not to give too many details
00:01:00 exactly on timeline on a variety of things.
00:01:03 But the vast majority of what I brought forward
00:01:06 came out in just the last few weeks.
00:01:09 So it was one of these things where I cared a lot
00:01:12 about these questions on platform design and algorithms
00:01:14 and how content censorship AIs work, that kind of thing.
00:01:19 In the last couple weeks,
00:01:21 that's when we started grabbing things
00:01:22 like the stuff around kids,
00:01:24 the stuff on teen mental health.
00:01:25 I knew that I had no business case to touch that. 0.91
00:01:28 Like I worked on foreign governments 0.66
00:01:30 interfering in the United States
00:01:32 and interfering around the world.
00:01:33 I had lots of reasonable business cases
00:01:36 to touch that information.
00:01:38 I never worked on kids' mental health
00:01:40 beyond like child predators.
00:01:43 And only then like as in the larger threat
00:01:46 intelligence framework.
00:01:48 And so when we started grabbing that kind of information,
00:01:50 I assumed Facebook would catch me.
00:01:52 That's when it happened.
00:01:53 That's when I assumed.
00:01:54 Well, they never caught me.
00:01:55 Like, we got all the way out the door before they ever realized, like, I had taken anything.
00:01:59 Ever get close?
00:02:00 Even the other girl that sat right across from you, she kind of had, was that scene where she comes and she tells you, here's the chip?
00:02:07 Did that ever happen?
00:02:08 That never happened.
00:02:08 That never happened.
00:02:09 So that's just movie.
00:02:10 That's just movie.
00:02:11 Okay, got it.
00:02:12 Were there other people involved internally trying to help?
00:02:16 No, just Jeff.
00:02:16 Just you.
00:02:17 Yeah.
00:02:17 And did anybody, like, you know, nowadays with HR softwares, you can, everybody wants to know sentiment.
00:02:24 They want to know, you know, like almost a minority report to find out we're about to lose Francis.
00:02:29 And based on email communication, you know, she's probably 90 days away from quitting and she's hit tier one crisis mode for the company.
00:02:36 Let's have a conversation with her.
00:02:37 Did anybody approach and say you seem very unhappy?
00:02:40 So I think I had that moment in my first six months at Facebook.
00:02:43 So when I was working on ethnic violence, like like more hands on when I was working on people targeting Americans with these targeted misinformation campaigns.
00:02:52 the level of seriousness
00:02:55 that the company brought to it
00:02:58 I was very concerned that not enough
00:02:59 was being done because remember I am watching
00:03:01 things like I have friends who work on
00:03:04 the ethnic violence team and they're getting
00:03:05 depression and having to talk about going on
00:03:07 medical leave like it's that bad
00:03:09 right people are feeling suicidal because like the company is
00:03:11 like the reality you learn inside
00:03:13 the company versus like what you can do
00:03:15 the gap is so big right
00:03:17 during that period of time
00:03:19 that's when I kind of clued in
00:03:21 that it was unacceptable to be unhappy inside the company.
00:03:25 They made it very clear that you were not allowed to be negative.
00:03:29 And so after about six months in, I realized,
00:03:31 oh, if I want to continue working on Facebook,
00:03:33 and that's around when COVID hit, and I was like,
00:03:35 oh, I don't want to lose my job.
00:03:37 You cannot visibly ever be unhappy with the company.
00:03:41 And so at that point, I kind of straightened up and flew right.
00:03:45 Like you were acting and you were being fake?
00:03:48 I would say I chose what to focus on, right?
00:03:52 And the good thing was that I had gotten shifted at that point,
00:03:54 because remember, they move you around constantly,
00:03:57 so you never get very deep in any one thing.
00:03:59 At that point, they had moved me on to modeling
00:04:02 how much influence anyone on the platform had.
00:04:05 So at that point, we were trying to identify who,
00:04:07 not just elected officials,
00:04:09 like let's say you are a labor organizer.
00:04:12 Can we identify that you have effectively power on Facebook?
00:04:16 Would you guys give a score to that person?
00:04:17 We would give a score to everyone.
00:04:18 Almost like back in the days, there was a website called Clout.
00:04:21 If you remember Clout with a K.
00:04:22 We kind of were calculating Clout scores for everyone.
00:04:24 And you have that on the back?
00:04:25 We made a prototype of it.
00:04:28 I got moved off onto threat intelligence towards the end of that summer.
00:04:33 So we made a prototype that was actually quite interesting.
00:04:36 And it's private.
00:04:37 It's yours, not public.
00:04:38 It's only for Facebook.
00:04:38 So we put it in place because Facebook treats influential people differently than it does average people.
00:04:44 In what way?
00:04:44 like, so one of the big revelations I brought out
00:04:47 was about a program called Crosscheck.
00:04:49 So Facebook had been public about the idea
00:04:50 that this thing called Crosscheck existed.
00:04:52 They had never disclosed the details
00:04:55 of like how you got on Crosscheck.
00:04:56 So you probably are on Crosscheck
00:04:58 given the amount of influence you have on the internet.
00:05:01 So if you're on Crosscheck,
00:05:03 if you're a celebrity, you're a big elected politician,
00:05:06 Facebook will not apply any of the content moderation tools
00:05:10 to your content. 0.96
00:05:11 You will end up in a human queue
00:05:13 that will take days.
00:05:15 So by the time someone actually looks at your content,
00:05:17 you have already gotten all the views
00:05:19 you're going to get probably.
00:05:20 80, 90% of the views you were going to get anyhow.
00:05:22 But no AI.
00:05:23 It takes 24 hours plus.
00:05:25 No, yeah, yeah, yeah.
00:05:25 Back then it was taking 73 days.
00:05:29 Yeah.
00:05:30 And this was a huge problem.
00:05:31 Like I'll give you an example.
00:05:32 There was a Brazilian soccer player
00:05:34 who sexually assaulted a woman
00:05:36 and she came out and accused him. 0.64
00:05:38 And he turned his fans on her.
00:05:40 like basically actively called for them to go after her and because he was in this special queue
00:05:45 it took days for them to get to his content and like she she got uh death threats she got people
00:05:51 show up at her home like all kinds of stuff happened because uh he he could call for violence
00:05:55 against her and no no by the time a human looked at the content it'd been days and days down the
00:06:00 road right so so there's this question which is facebook has these different programs in place
00:06:04 to protect, say, a local politician, a state-level politician.
00:06:09 If we don't have data sources that can tell us this is who needs protection,
00:06:14 we have to figure out on our own who needs protection.
00:06:16 Because some of the other things are things like making sure you have an extra recovery phone number
00:06:21 or, like, we make sure you turn on two-factor authentication because you're more likely to get hacked.
00:06:25 You know, the intention of that project was to make sure that the people who needed protection got protection.
00:06:31 So on the back end, did you have the highest cross-check score on influence for people?
00:06:36 Was Trump number one?
00:06:37 Was Elon ranked?
00:06:38 So my focus was not on people we already had data on.
00:06:42 So Trump would already have been under one of our major federal politicians.
00:06:47 There were like 35 different lists that might be how you'd get cross-check protection.
00:06:51 If I'm in a room, I could see the ranking of who's the most influential to the least?
00:06:55 Yeah.
00:06:56 For the prototype that we built.
00:06:58 And you probably see it moving like this, right?
00:07:00 But we only focused, everyone who already had protection, we filtered them out because we're like, oh, you're already protected.
00:07:05 So we focused on everyone beyond that.
00:07:08 And so, for example, I remember one of the people in our top 10 was a National Medal of Freedom winner.
00:07:16 Like she's gone and returned more than like a million acres of Native American land.
00:07:23 She's very, very influential, but she is not an elected official. 0.98
00:07:26 those are the kinds of people who we didn't know about until we 1.00
00:07:30 mined them for influence did you reach out to them or you just watched them
00:07:34 they didn't know you're doing this you're just watching them so my specific job was just to find them
00:07:38 because there are other parts of the company that provide them with protection
00:07:41 because I'm just a data scientist figuring out like can we find the people with influence
00:07:46 so that then the parts of the company that give them special treatment can give them special treatment
00:07:50 because I remember something happened with Facebook where guys who were making 30 40 50 100
00:07:54 thousand dollars a month with facebook revenue that was coming and all of a sudden it split
00:07:58 yeah and you know to a fifth and a lot of people that were relying on that income they lost it i
00:08:03 don't know what the event was so that that's a that's a huge deal like most um online news
00:08:08 sources had this happen to them so you remember this one this happened uh maybe 2021 2022 right
00:08:14 basically facebook came and said we are no longer going to uh prioritize the news and so all these
00:08:19 online news sites suddenly lost all their traffic and revenue and revenue yeah yeah traffic is
00:08:24 revenue was that intentional i think facebook um one they knew that they were facing in some
00:08:30 countries like canada for example um canada has put in place laws that say hey facebook like your
00:08:34 business is built on distributing the content of of people who who who didn't pay like if you want
00:08:42 there to be a news ecosystem you have to give at least a little bit of funding to the news sector
00:08:46 I think Facebook looked at them, looked at the Australians, and basically said,
00:08:50 we're just going to distribute less news because we don't want to risk having
00:08:53 governmental interference in the future.
00:08:56 So when you were going through blowing the whistle on Facebook
00:09:02 and you're dealing with Jeff, Jeff, you know, he's a known guy in this space.
00:09:07 I hope I'm saying his name correctly.
00:09:08 He's much more competent than with the movie. 0.95
00:09:09 They made him seem like he's a dummy.
00:09:11 I've got to defend him. 0.97
00:09:11 I've got to defend him.
00:09:12 Okay, that's good to know because I was not convinced you're going to be 0.99
00:09:15 that guy and be as dumb as they made him out to be not dumb but it's like oblivious yeah that's 0.97
00:09:20 what i'm saying so he knows what he's doing so while you're going was it did he at all tilt you 0.99
00:09:26 did he at all influence you to go like did he say something like hey you're about to change the
00:09:31 world you're going to be on everything do you want to take this or was he kind of like encouraging
00:09:36 discouraging you sure you want to do this what role did he play so um i never intended to come
00:09:41 forward publicly so like that decision was made like six weeks before the first wall street journal
00:09:46 article maybe six weeks maybe four weeks before the first wall street journal article came out
00:09:50 um so he definitely um he definitely uh counseled me against coming forward publicly but like when
00:09:59 we were in the thick of it um i like i never really wavered on that i was going to provide
00:10:06 him with materials the question was like were we like when we started out like i think i probably
00:10:11 never expected to work with him for more than a month or two um and like i never thought we would
00:10:16 bring out 20 000 pages of documents right like i never thought we would end up being the reason why
00:10:22 47 states sued and won against facebook you know like i never saw this future in front of us um
00:10:29 but he uh was really one of the things he told me that i was very very grateful for
00:10:34 and i've always carried with me is he's always said you can stop anytime you want to like you
00:10:39 owe no one anything like this is not you you're like at some point it's not fun anymore don't
00:10:45 like you don't have to be a public whistle so that happened yeah that happened that was there
00:10:49 ever a flirtatious moment between the two of you or no he's he's married oh he's married i didn't
00:10:54 know that and and the thing that i think is so sad about the movie um is so the place we actually
00:10:58 met so we didn't he did not live in a hotel he lives in oakland um his wife is a garden designer
00:11:02 and we the place we actually met most often um in san francisco was um uh his wife has built this
00:11:09 glorious garden like they they rescue cactuses like they go down to like um like joshua tree
00:11:15 and like like people will say like i'm going to cut this cactus down from my front yard because
00:11:18 anyone wanted they will go and rescue cactuses so they have like these giant cacti and like
00:11:23 uh trellises with flowers and it would never look realistic on camera but like that's where we met
00:11:29 Rescue cactuses?
00:11:30 Yes, you can rescue a cactus.
00:11:31 How do you, because it's not going away.
00:11:33 How do you rescue?
00:11:33 So like if a cactus like lays on the ground, it can like put down roots.
00:11:38 So you can like take an arm off a cactus and replant it.
00:11:41 It's wild.
00:11:42 Yeah, it's pretty wild.
00:11:43 Yeah.
00:11:43 So imagine like the place that, so it's like cold in San Francisco.
00:11:47 It's like dreary, foggy, and you get to go to a nice sunny Oakland.
00:11:50 Because remember, the rich used to have their summer homes in Oakland because it's sunny and more pleasant.
00:11:54 I didn't know that.
00:11:54 Yeah.
00:11:55 So, you know, and so that's where we met.
00:11:58 Wow.
00:11:58 okay and so the when when you're at a point you're like i'm going public with this and
00:12:05 you know i'm gonna go and i'm gonna be on camera and they're gonna see my face they're gonna know
00:12:10 who francis hagan i'm gonna be doing that did you get any weird calls did you get any weird
00:12:16 visits did you get any weird texts did you get any weird people wanting to talk to you or no
00:12:20 so no one knew who i was until the day of 60 minutes right so we um the only people who knew
00:12:26 who i was was was like the wall street journal so they knew my name um and the that that scene
00:12:32 where the like the francis howton character comes in and like meets all the wall street journal 0.99
00:12:35 people like that actually happened that did happen um i the uh my my um my moniker lady as mean as 0.79
00:12:42 she was so so his actual editor is a dude um and and i actually was he a jerk no he's wonderful
00:12:47 okay um so you just being diplomatic no i'm i'm like like i i was i was telling jeff like i thought
00:12:52 it was the a little sad because like his editor was like as much like a part of the team almost
00:12:56 right because like we'd have to call him and defend what we were doing right because like
00:12:59 we'd call him up and be like we found this thing it's you know and and the job of an editor is to
00:13:03 be skeptical right and the job the editors be like did you really is that really true like how
00:13:08 can you prove that is that and so um having someone who is skeptical and is like the wall
00:13:13 you bounce things off of is how you have better ideas and so i i i feel like the that role is
00:13:18 missing from the movie interesting and and your experience now with wall street journal leaving
00:13:23 them what's your level of respect and credibility you have for their approach and how they do things
00:13:28 i thought they did a really good job um the the one thing where um i think they um i don't know
00:13:35 what happened on the inside um but i think they got a bunch of pressure because um there were like
00:13:42 11 stories all done ready to go and after the first say week came out they were like oh we
00:13:49 can't publish anything more for a while like one of the big places of conflict was um jeff had
00:13:55 he had a team of like four journalists who worked with them and they had fleshed out these giant
00:14:00 stories that were out of the files and by the time we had to open things up to the rest of the press
00:14:05 consortium um he was he was like chomping at the bit like he wanted to be able to publish these
00:14:10 stories and i don't know what's happening inside the journal that they held him back because he
00:14:15 could have been the one who broke more big stories than he did but they didn't let him actually push
00:14:20 out the door until sometimes weeks or months later some of those stories was there anything
00:14:24 that was left out it was left out i mean the fact that the press consortium published lots more
00:14:30 stories um there is all this question of scale right so the wall street journal has a certain
00:14:34 level of staffing and they have a certain focus and so there's going to be stories that are
00:14:38 relevant, say, only in Poland.
00:14:40 So, for example, there was a couple of stories around politically influential people that
00:14:47 came out of some of these cross-check documents that those stories are not going to get published
00:14:52 by the Wall Street Journal because they don't care about Polish politicians.
00:14:55 But the paper out of Warsaw is going to publish them.
00:14:59 Got it.
00:15:00 So, now, they did come out.
00:15:02 I think when that was going on, Facebook came out and says, you weren't a senior executive.
00:15:05 You worked there for less than two years.
00:15:06 You never attended decision-making.
00:15:08 meeting with c-suite executives you acknowledge it during your congressional hearing zuck
00:15:12 said your allegations were false i think he responded publicly on october 5th 2021
00:15:17 he rejected the allegation that facebook knowingly prioritized profits over user safety
00:15:22 then facebook said she wasn't an expert on instagram's effects on children andy stone
00:15:27 emphasized that hagen uh hadn't personally worked on instagram child safety research
00:15:31 zuck said most of us just don't recognize the false picture of the company that is being painted
00:15:37 His response was that Francis, in interpretation,
00:15:39 failed to acknowledge the company's extensive investment in safety and research.
00:15:43 Zuck says your accusations unfairly characterized thousands of employees
00:15:46 who genuinely care about protecting users.
00:15:50 These are some of the things that he took.
00:15:52 Do you agree with him on what he said?
00:15:53 What I find so funny about the, like, we don't agree with this characterization thing,
00:15:57 is one of the things that came out from the lawsuits against the company about Instagram
00:16:01 was an email from Adam Masseri to Mark Zuckerberg
00:16:04 and a couple of other C-suite folks
00:16:06 about Instagram and child safety.
00:16:09 They came out, the email happened like one or two weeks
00:16:12 before the first story was published
00:16:15 in the Wall Street Journal.
00:16:16 And the reason why this timing is so interesting
00:16:17 is we gave them three weeks to do comment, right?
00:16:21 Like the Wall Street Journal was cautious enough,
00:16:23 impaired enough that they didn't give them
00:16:25 three days to do comment.
00:16:26 They gave them three weeks.
00:16:28 So sometime between when they were tipped off
00:16:29 that these stories about kids and Instagram were coming.
00:16:32 A week or two later, Adam Seri writes to Mark and says,
00:16:35 in order for us to actually do what we claim we're doing publicly to protect kids,
00:16:40 we need another 45 people working on kids.
00:16:44 And the CFO writes back, there's no money to do that.
00:16:48 This is documented.
00:16:49 This is documented.
00:16:50 This is an email that came out through the court cases.
00:16:52 And they have since claimed, oh, no, no, no, no, we actually did fund that later on.
00:16:56 But like the email thread that we have,
00:16:57 Like we don't, allegedly how all this got funded is via verbal commands.
00:17:03 There's no paper trail saying this ever did get funded.
00:17:05 But Adam and Sarah said, in order just to do what we say we're doing publicly, it would take 45 more people.
00:17:16 Handcrafted across five Italian factories.
00:17:19 With builder technology underfoot, it may be the most comfortable shoe you will ever own.
00:17:25 Introducing the all-new FLB shoe
00:17:28 Finished in new chocolate caretza leather
00:17:32 Made to move
00:17:33 Built to endure
00:17:35 Designed to impress
00:17:37 If you enjoyed this video
00:17:40 If you want to watch more videos like this, click here
00:17:42 And if you want to watch the entire podcast, click here