Attorney General Raúl Torrez Marks Conclusion of Final Phase in Landmark Trial Against Meta; Read Testimony Highlights

Santa Fe, NM — Attorney General Raúl Torrez released the following statement as the New Mexico Department of Justice (NMDOJ) heads into the conclusion of the final phase of its landmark child safety lawsuit against Meta today. Phase two of the trial, which followed New Mexico’s historic victory against Meta in March, is expected to conclude today with Meta’s final witnesses.

“Two months ago, a jury of New Mexicans found that Meta’s product design intentionally harms children and that Meta lied to parents about the safety of its products,” said Attorney General Torrez. “New Mexico’s victory, and the $375 million in civil penalties, changed the legal landscape for Big Tech, establishing that they can – and will – be held accountable when they violate laws that protect kids. It is time for Mark Zuckerberg to finally prioritize child safety, and I am looking forward to hearing the judge’s ruling on the changes Meta must make to protect children in New Mexico.”  

During phase two of the trial, the NMDOJ argued that Meta’s platforms constitute a public nuisance and sought injunctive relief that would require Meta to fund its share of the State’s abatement plan to address teen mental health needs attributable to social media.

In addition, the State is requesting that Meta make specific changes to its platforms/company operations in New Mexico to protect children. This proposed injunctive relief includes:  

• Independent oversight through a court-appointed child safety monitor; 
• Effective age verification to prevent adults from posing as minors, ensure that all teens receive appropriate safeguards, and enforce minimum age requirements for pre-teens; 
• Safer algorithms for children that do not prioritize engagement over well-being; 
• Remove addictive features for children like infinite scroll, autoplay and push notifications during school and sleep hours; 
• Restrictions on end-to-end encryption in private communications with minors to prevent predators from operating in secrecy; 
• Prominent warning labels about risks of Meta’s platforms to teens;  
• Permit New Mexico law enforcement to operate undercover accounts to test child safety on Meta products; 
• Prevent any adult not directly connected to a user under 18 from messaging that user; and 
• Permanent bans for adults who engage in or facilitate child exploitation. 

The State is seeking an abatement plan that requires Meta to fund its share of a plan to address teen mental health needs attributable to social media, or $779.5 million.

In March, New Mexico became the first state in the nation to prevail at trial against a major technology company for misleading consumers and endangering children after a jury found Meta liable under the Unfair Practices Act and ordered the company to pay $375 million in civil penalties. 

A factsheet on the changes to Meta’s products sought by the NMDOJ is available HERE. 

Below are key themes and highlights from expert witness testimony throughout phase two of the State’s trial vs Meta: 

Clarie Miller, New Mexico Department of Health Lead Suicide Prevention Coordinator, on the Link Between Social Media and Suicide 

Topic: The Limits of Individual Prevention Tools and the Need for Product Changes   

Question: “In your experience as suicide prevention coordinator, have you found that teens’ use of these tools alone is sufficient to prevent the harms we discussed?” 

Miller: “No. […] It’s not enough. There are hazards of the product that no one else has control of. So even though I can offer them these tools, I didn’t create the product.”  

Question: “Based on your experience and observations, what would be helpful for New Mexico’s suicide prevention efforts?”   

Miller: “More of everything we’re doing, more trainings, more funding, more resources in the community, more access to healthcare in our rural areas.” 

---

---

—–  

John Talley, NMDOJ’s Internet Crimes Against Children Special Agent in Charge, on Meta’s Failures Reporting and Preventing Online Child Sexual Exploitation 

Topic: The Challenges Meta Creates for Law Enforcement Investigating Internet Crimes Against Children 

Question: “Are Meta’s…in your experience, are Meta’s subpoena returns easy to use?” 

Talley: “Absolutely not, no, sir. […] The returns we receive from Meta typically can sometimes be thousands of pages and it’s a bunch of – it looks like one long run-on sentence, basically. One of my agents just received almost 4,000 pages, and there’s no section breaks that tell you what you’re looking at, conversations don’t necessarily say who’s speaking and what’s going on, so we have to try to decipher those best as we can.”  

Question: “Are you able to compare the quality of Meta’s CyberTips with other platforms?”   

Talley: “We do, yes, sir. The majority of the Meta tips are not necessarily actionable by New Mexico State law standards. There’s a lot of meme tips – what we call junk tips – like I said, we really can’t act on.”  

—–  

Topic: The Failures of Teen Accounts 

Question: “In your personal experience, have you observed whether children and especially teenagers lie about their age when signing up for Instagram?”   

Talley: “I have actually never observed a child using a Teen Account. I’ve seen my own daughter’s account which she lied about her age to have an Instagram.”  

—–  

Nathaniel Lubin, Public Health Abatement Expert, on the Importance of a Court-Appointed Child Safety Monitor
  

Topic:  The Responsibilities of the Monitor 

Lubin: “[…] there’s a long history of Meta making representations that subsequently were not, in fact, carried out in terms of practice. And so a monitor is appropriate to ensure that those mitigation efforts actually occur.” 

—–  

Topic: The Monitor Reviewing Internal Experiment Data, Meta’s Safety Claims 

Lubin: “So the monitor would have access to the internal experiment data along the lines of what this is talking about here, as well as product experiment data. […] So they could directly observe whether these kinds of questions are being asked. And in cases where they’re not and think that it needs to be, they have the right to request that data be collected.” 

Lubin: “Meta has very repeatedly and publicly in its advertisements and elsewhere, talked about the value of these kinds of safety features, teen accounts and etc., that are predicated foundationally on accurate age verification, and so the monitor being able to report on how accurate those claims actually are is central.” 

—–  

Topic: On Meta’s Failure to Enforce its Policies 

Lubin (reading an internal Meta e-mail from 2020): “It says, ‘Glad to hear the team is thinking more re device blocking. […] Number one, perm block actually just one year for terrorism, CEI, CEI solicitation, revenge porn, content-based extortion, and sextortion. Temporary block two weeks for nudity, bullying, NAZI, drug, gang, hate speech, cartel, graphic violence, and impersonation.’ And then ‘Do not block for self-harm promotion, eating disorder promotion, copyright, trademark, counterfeit, spam, U13, sex offender, et cetera.’”

Lubin (reading an internal Meta audit from 2023): “Yes. I think this is what was cited in an expert report, but it says, ‘The WhatsApp Central Integrity team does not always ban community users with three or more counterterrorism strikes within 90 days per WhatsApp internal enforcement protocol. As a result, potential bad actors may proliferate on the platform. For example, on the sample date approximately 200 community creators who qualified for user ban were not banned.’” 

—–  

Topic: On the Suppression of Child Safety Research 

Question: “And how did this [internal chat between Meta employees] – and let’s actually refer to specific language within this chat. If you can look at – I’m going to only say the first name Anya, the OC line. Can you start reading that?”

Lubin“Yeah, so ‘OC wants me to remove language in my research docs that suggests that teens are particularly vulnerable, e.g., quote, period of particular developmental sensitivity, because they are concerned the FTC and others will use that language to say that Meta knew these populations were at heightened risk, and, therefore, we should have design product features to avoid harming them’ and she says ‘And I’m like yes.’” 

Question: “And keep going and then there’s an explanation of outside counsel. Can you turn to Tom’s line 2 down –” 

Lubin: “So Tom says, ‘Fairly sad state of affairs but familiar territory.’” 

Question: “Okay. So what was the significance of this document to your opinion about the monitor’s role with respect to research?”

Lubin: “So the goal of the monitor, as I said before, is to provide independent assessment to ensure that research is being understood and implemented so as to assess whether harms are continuing. And so this document, as well as the others we’ve been talking about, suggests that absent oversight from the monitor, there’s significant risk that because of counsel or others inside the company, research may be modified or prevented that would be able to do that kind of assessment. 

Question: “If we can look at one line on page 2. On the very top of the page, Mr. Lubin, do you see the chain from Adam Mosseri? […] And can you read the first line of that e-mail?” 

Lubin (reading a 2017 e-mail where Meta employees discuss excessive usage and data): “It says, ‘On the numbers 1.3 percent of people in the US spending 30-plus hours a week sounds like a lot.’” 

—–  

Fallon McNulty, Executive Director, Exploited Children Division at the National Center for Missing & Exploited Children (NCMEC), on Fixing Meta’s Reporting Failures  

Topic: The Limits of User-Based Reporting  

Question: “And what would you highlight as what Meta could be doing that it’s not doing now?” 

McNulty: “One of the largest pieces of feedback that NCMEC has provided in years past is certainly a reliance on user-based reporting. That primarily would speak to Facebook Messenger and the implementation of end-to-end encryption, where now there is not the ability for child sexual exploitation to necessarily be proactively identified or disrupted.”

—–  

Topic: Discussing a March 16, 2026, Letter NCMEC Sent to Sen. Grassley in Response to His Letter 

Question: “Okay. On the same page, does the letter raise issues about Meta’s adult classifier and problems with false reporting?”

McNulty: “Yes, it does […] NCMEC’s understanding of the adult classifier is an AI tool that assists with age estimation […] And that information was being used, in turn, to generate CyberTipline reports. The feedback that we had received from law enforcement was that on multiple occasions, after they independently reviewed the reports, they found that it actually was two adults who were in communication with one another.”

Question: “And what does this –  the problems you’ve seen with the adult classifier, what does that say to you or what does that indicate to you about Meta’s testing of its classifiers?”

McNulty: “[…]our understanding, based on the CyberTipline reports that we receive and the amount that have an indication that this classifier is being used, seems that it was very quick to launch, and there was not testing necessarily that the CyberTipline saw or was able to provide feedback on.” 

—–  

Topic: How Law Enforcement Feel About Meta’s CyberTips 

Question: “Do you continue to see a reluctance to trust CyberTips from Meta on the part of law enforcement?”  

McNulty: “We do currently know of multiple law enforcement agencies who have indicated to NCMEC that they do not wish to receive reports from Meta based on quality issues within those reports.” 

—– 

Topic: The Technology Exists for Meta to Better Identify and Report Novel CSAM 

Question: “In your experience, are there platforms, other than Meta’s platforms, that are better at identifying and reporting novel CSAM to NCMEC?”

McNulty: “Yes, there are platforms that come to mind.”

Question: “And are you aware of success rates in terms of the platforms that do the best job at detecting novel CSAM?”

McNulty: “So one of the platforms that we frequently see novel CSAM reported from would include Google. […] they do include in some of their publicly available transparency reporting the rate at which they proactively detect CSAM material versus what has been viewed on the platform by users.”

Question: “And do you know what that rate is?” 

McNulty: “It is over 99 percent.”

—– 

Dr. Brian Levine, Director, UMass Cybersecurity Institute, on Age Verification Capabilities and End-to-End Encryption 

Topic: On How Age Verification Technology Exists for Children Under 13 

Question: “Just to be clear, Dr. Levine, do you believe, given technology existing now, like Yoti, and age verification, that 99 percent for under 13s is technologically feasible?”  

Levine: “Yes. […] Part of my opinion is based on what I’m aware of already is available on the market. Yoti – as I believe I said this yesterday, Yoti has announced that their products are exactly that, 99 percent accurate for detecting that someone 6 to 12 is, in fact, under 13. Not to repeat myself but I’m gonna – this is a product that Meta already uses – or a partner that Meta already uses [for Facebook dating].”  

Question: “Would age assurance have an impact on child safety on Meta’s platforms in your opinion?”  

Levine: “[…] I’ve said the Teen Accounts and other mechanisms they have in place are dependent of knowing the person’s age. So, you can’t protect kids if you don’t know that they’re kids and you can’t protect them from adults if you don’t know that adults are adults.”  

—–  

Topic: On the Dangers End-to-End Encryption Poses for Young People

Levine: “I mean, these direct messages are where people exchange CSAM, where images of nudity of young teens can be exchanged. There’s nudify apps out there. This is very dangerous. There’s few places in society where we believe children who are, especially even just 13, should be talking to adults in an unmonitored sort of way.” 

—–  

Zachary Ward, Assistant Professor of Health Decision Science at Harvard TH Chan School of Public Health, on Mental Health Harms Tied to Social Media Use 

Topic: On Mental Health Harms New Mexico Youth Experience as a Result of Social Media 

In his testimony, Dr. Ward reviewed slides outlining his analysis that social media is a substantial contributing factor to mental health harms among youth in New Mexico. His analysis showed that one in ten youths in New Mexico has a mental health concern that is attributable to social media. He looked specifically at evidence linking social media with depression, anxiety, sleep deprivation, suicide risk factors, and disordered eating.  

Question: [reviewing a slide with a chart of Dr. Ward’s findings] “And so what does the 21% reflect?”

Ward: “So that’s averaged across all of these harms. So that’s saying that 21% of youth in New Mexico who have any of these harms have it because of social media.” 

Question: “And can you read the bottom line?” 

Ward: “The bottom line on the slide, so this is just looking at the same data in a different way. So this says that one in ten youths in New Mexico have a mental health concern that is attributable to social media.” 

Ward: “So, when you combine across middle school and high school boys and girls, 22 percent of youth who have at least one risk factor for suicide would not qualify – would not have those risk factors but for their social media exposure.”  

Ward: “[…]when you look at all youth in middle school and high school in New Mexico […] 15 percent of them would not be experiencing sleep deprivation but for their social media exposure.”  

Ward: “So, 45% of girls who are experiencing disordered eating behaviors would not have had those problems but for their social media exposure.” 

—–  

Michel Protti, Chief Privacy and Compliance Officer for Product at Meta, Concessions During Testimony 

Topic: The Feasibility of Requiring Parental Consent for All Under-18 Users to Opt Out of ‘Private By Default Settings’ 

Protti: “‘A’ here, ‘Private By Default settings,’ is essentially on – all under-18s are defaulted into private accounts, but status quo, 16- and 17-year-olds are able to move themselves into public accounts without parent consent. 15-year-olds and below require parent consent. We would propose extending that parental consent requirement to all under-18s. It’s not something we have already done […] but given the State’s demands, out of deference to the Court here, where this is something we’d be willing to consider.”

—–  

Topic: The Feasibility of Blocking Push Notifications During School Hours 

Protti: “Yeah. So we’d also be willing to input by default an 8 a.m.-to-3 p.m. block on notifications during broadly school hours here. With some exceptions, we think messaging notifications should come through if it’s a parent trying to get in touch with a kid, security, and integrity as well. But, yeah, we’d be willing to extend that.”

—–  

Topic: Admitting the Court-Appointed Child Safety Monitor Would Ensure Real Age Verification  

Question: “Would the monitor also have oversight of the proposed U13 prediction model?” 

Protti: “[…] But the monitor should be able to ensure that we have the age assurance elements in place that the State has ordered.” 

 —–  

Topic: Admitting that Meta Has the Technological Capability to Detect Under-13 Users 

Protti: “So what we can do for under 13s, though, is instead of running this always-on predictive model on everyone, what we can do is look for these signals in their content, which are sort of warning signs that this person may be under 13.” 

—–  

Topic: Examining Evidence on Meta’s Plans to Implement – and then Withdraw – Age Verification in Arkansas Once Legal Pressure Lifted   

Question: “All right. Mr. Protti, this chat starts from a message from you on August 30 that starts ‘You around? Any chance I can call you for 5 mins, re: Arkansas’; is that correct? […] And a few lines down, there’s another message […] And you write, ‘My take is if we don’t hear anything from the Court, we turn on our compliance plan as planned. If when the Court rules, there’s a stay on State enforcement, we should pull back our compliance plan and assess the landscape. We’d then monitor the landscape and legal  developments to decide if/when we turn our compliance plan back on.’ Is that accurate?” 

Protti: “Yes”   

Question: “Okay. And so as you testified – or  testified earlier, you’d agree that as of the date of  this message, Meta was in a position to comply with the Arkansas requirement; correct?”

Protti: “Yes”   

Question:  “And it was technically feasible as of that date; correct?”    

Protti: “It was.”   

Question: “And Meta never considered  – ”   

Protti: “I should interject on this whole line of questioning, the State’s proposals around age verification, parental consent, are technically feasible […] So, can we implement them? Yes.”   

Question: “Okay. Let’s read one more line from this chat. Antigone Davis responded to what you just read, “If you mean the Court rules against the injunction, then I remain okay with plan of record, but generally have worries about compliance requirements outpacing our ability to push for OS solution”; correct?”    

Protti: “Yep.”  

Question: “And Ms. Davis was expressing concern that turning on compliance in Arkansas would serve as a proof point that Meta could do this and didn’t need the App Stores; right?”