AU, Mr Ben, Head, Public Policy, Australia and New Zealand, Snap Aus Pty Ltd [by video link] GARLICK, Ms Mia, Regional Director, Policy, Meta [by video link] LOFSTEDT, Ms Jessika, Head, Public Policy, Australia and New Zealand, TikTok [by video link] LORD, Ms Rachel, Senior Manager, YouTube Government Affairs and Public Policy, Google [by video link] CHAIR: Welcome. I understand that information on parliamentary privilege and the protection of witnesses giving evidence to Senate committees has been provided to you. Yes? Thank you. You may have heard me say earlier that we're running a little bit behind time, and I apologise for that. I will invite you each to make a short opening statement, but, if you wouldn't mind, to keep them relatively brief, and then if you want to add anything further you can put that on notice or table it at the conclusion. At the end of your opening statements I'll invite members of the committee to ask questions. Ms Garlick : Thank you for the opportunity to appear before you today. I want to say at the outset that Meta shares the Australian government's and the parliament's commitment to provide young people with age-appropriate experiences online, and this is something we've been investing in for a number of years. Most recently, we launched Teen Accounts in Australia, in September 2024, after many years of research. That represents the most significant change we've made to young people's experiences on our platform. It has stronger privacy settings, stronger content controls and parental oversight built in by default. When it comes to compliance with Australia's social media minimum age legislation, I want to be very clear that Meta took this seriously from day one, and whilst we believe that legislative frameworks that incentivise industry to invest in age-appropriate experiences are preferable to blanket bans, our approach has been to disagree but to commit. We were the first major platform to begin notifying and removing underage accounts before the law took effect on 10 December last year. By 11 December, we had removed access to almost 550,000 accounts, and that enforcement has continued every day since. We've been reporting our progress regularly to the eSafety Commissioner and we've now removed more than 750,000 accounts across Facebook and Instagram in Australia for people that are under 16. For us, building out an age-appropriate experience is an iterative process. We've adopted a waterfall approach, which is a data minimisation approach consistent with the regulatory guidance from the eSafety Commissioner and the findings of the age-assurance technology trial final report. Consistent with that, we've also started rolling out AI-powered detection, both text and visual, to proactively identify underage users. We've simplified user reporting. We've augmented human review with LLM based systems, and we've implemented antirecidivism measures to prevent young people from simply reregistering. Enforcement is only part of the job. Between June and July this year, we ran an education campaign across Facebook and Instagram in Australia that reached 1.3 million parents, pointing them to practical guidance on how we detect and remove under-16 accounts, what happens to a teen's account and memories, and how to check and correct stated age. We think this is really important because we need to keep parents informed as part of making these protections work. Outside of our work to comply with Australia's social media age ban law, we continue to invest in and advocate for industry-wide whole-of-ecosystem age-appropriate experiences. We're a founding partner of the OpenAge Initiative, which is a non-profit focused on interoperable privacy preserving age assurance, and we're working to test the launch of AgeKey as part of this initiative for Australian Facebook users later this year. So this is another step forward in reducing the burden on individuals while giving platforms stronger signal to ensure compliance. I hope that this framing is helpful, and I'm looking forward to the discussion. CHAIR: Google. Ms Lord : We don't have an opening statement today. CHAIR: TikTok Australia. Ms Lofstedt : I'll keep it brief. Thank you to you and the committee for the invitation to appear today. TikTok is a global discovery platform where more than 10.5 million Australians come to connect, share their passions and find inspiration. Our top priority is keeping our communities safe and we share the government's policy objective to ensure people have safe, age-appropriate online experiences. We take our compliance obligation under all Australian law seriously. Since December last year, a key plank of our online safety work has involved supporting parents and their teens through the changes introduced when the government's new social media minimum age law came into effect. As you're all aware, these laws were a world first. Both the government and the eSafety Commissioner recognised this and have acknowledged that its implementation needs to be iterative. At TikTok, we agree, and we continue to work closely with the Office of the eSafety Commissioner and the government, sharing how we're implementing and complying with the law. We strive for our engagement to be grounded in cooperation and continuous improvement. You should all have a copy of the submission TikTok made to this committee, which sets out our approach to the law. I'm here today to help answer your questions, so let me conclude by saying we take our responsibility to protect our community incredibly seriously. It's the most important thing we do. CHAIR: And Snap. Mr Au : Thank you, Chair, and committee members. Snap respects the critical role this committee plays in scrutinising policy and informing public debate. We have appeared before you on multiple occasions to share our perspectives and to answer your questions, and we remain committed to assisting your inquiry today. While Snap does not agree with eSafety's view that Snapchat is within scope of the social media minimum age, Snap is acting consistently with the law and taking the necessary steps to meet the requirements in the legislation. This includes a range of measures to prevent under-16s in Australia from holding accounts with Snapchat, including the use of facial age estimation, where a user takes a selfie which is then processed on device; ConnectID, where age is verified through a connection to an Australian bank account; and AgeKey, which is a reusable token users can create by verifying their age and submitting a government photo ID. While it is an unusual situation to testify alongside an active eSafety regulatory process, we welcome this discussion. We will endeavour to be as helpful as possible to the committee today while ensuring we remain respectful of the commissioner's ongoing work. I welcome your questions. CHAIR: Thank you very much. I'm going to go first to Senator Ananda-Rajah. Senator ANANDA-RAJAH: This is to all the platforms. In February, Apple announced that it had released its declared age range API, which could share age signals with platforms to help determine under-16 users. Despite this, I understand that not all platforms have opted into this. I'd like an explanation as to why, and I might start with Meta. Ms Garlick : I think it's an excellent step that Apple has taken because it demonstrates the way in which we can have a whole-of-ecosystem age-assurance approach, which is certainly something that a number of people within the industry have been advocating for. We're looking at different signals that we can be ingesting, and we hope to have more to share on that soon. Senator ANANDA-RAJAH: Does that mean you haven't opted into this? Is that right? Ms Garlick : We're working on it, and we've started the conversations to allow that to commence. But I don't think that that's been part of our suite to date. We've adopted other proactive detection signals, in part because the strength of the signal is very weak. It's actually less strong than stated age. We're working on a range of different signals. Senator ANANDA-RAJAH: Do you have a timeline as to when you would consider actually adopting this? Ms Garlick : I don't have anything I can share with you now, but I'm happy to come back to you with details if that's helpful. Senator ANANDA-RAJAH: That would be helpful. The others can go ahead. Ms Lord : At Google, we have agreed to start ingesting the Apple API signal. That will be rolling out imminently as far as I understand it. I should also flag that we have made a similar play API signal available to developers as well. Senator ANANDA-RAJAH: TikTok? Ms Lofstedt : This is a signal that we are currently actively testing. We welcome the availability of the signal from Apple and are actively testing it at the moment. Senator ANANDA-RAJAH: Do you have a timeline as to when you will deploy it or adopt it? Ms Lofstedt : As you can appreciate, with this type of signal, we want to obviously verify that it actually helps in the overall process. That's a current evaluation. Senator ANANDA-RAJAH: Snap? Mr Au : I can confirm that Snap has been actively testing the Apple declared age range API. I don't currently have a timeframe, but it is actively being tested. Senator ANANDA-RAJAH: This is to Meta. What steps have Facebook and Instagram taken to identify users who may have falsely declared their age? How do you test whether those steps are effective? Ms Garlick : We've got a range of different measures that we take. One signal that we have is stated age. We also have enforcement checks so that people can't create accounts under the age of 16 in Australia. We've also started rolling out proactive detection on text and visual signals. We also have been improving and augmenting our user reporting flow and responding to user reports as well. We have a range of antirecidivism measures. Senator ANANDA-RAJAH: I'm going to stop you there. I'm sorry. That sounds a bit jargony. Can we strip back some of the jargon. Could you describe to us some of those measures in plain language. Ms Garlick : Basically, we're trying to stop people from being able to create an account that is under 16. We're using measures to try to identify whether people's stated age and what we suspect to be their actual age are different. We're then taking proactive steps to remove those accounts. We're also trying to make sure that people can't try to create an account if we have removed it. There are a range of different measures that we're taking. If someone is, for example, changing their birth date—we were one of the first platforms to start giving people notice that they would lose their account. If they suddenly change their birth date to be over 16, we put them in an age check where we then ask them to confirm their age either through ID or facial estimation. So there are a range of different measures that we're taking to comply. We rolled these things out gradually in the lead-up to the law, and we've been rolling them out gradually since the law so that we can continue to test and iterate on them to check their effectiveness. Senator ANANDA-RAJAH: Have you got any data on how many young people try to re-join the platform? Ms Garlick : I don't think we have specific data on that. We've certainly disclosed that we've removed over 500,000 accounts in the lead-up to the law taking effect, and now the total number of accounts, as of today's hearing, is over 750,000 accounts. So, wherever we're aware of someone that we suspect to be under 16, we're removing those accounts. Senator ANANDA-RAJAH: And what's the denominator? Just remind us. How many children, or users under the age of 16, do we believe are using your platform? Ms Garlick : I don't have that to hand, I'm sorry. That might be a question for the eSafety Commissioner. Senator ANANDA-RAJAH: You don't know? So you've removed 750,000, but you don't know what the denominator is? Ms Garlick : From our perspective, if we suspect someone is under 16, we're removing that account. Senator ANANDA-RAJAH: But you don't have an estimate? You don't have any sense of how many people, children, are on your platform who are in that age range—not even an estimate? One million? Two million? Any idea? Ms Garlick : No, I'm sorry. Senator ANANDA-RAJAH: There's been a disturbing report from the US of Meta shredding evidence in regard to the Andrew Forrest trial. Have you shredded evidence, or destroyed any evidence, in the period that we've been debating this amendment? Ms Garlick : Absolutely not. We certainly take our legal and ethical obligations very seriously, and I'm grateful for the opportunity to clarify that for you today. Senator ANANDA-RAJAH: That is reassuring. To TikTok: how do you identify underage users who access your service through repeat account creation, altered birth dates, VPN use or other circumvention methods? Ms Lofstedt : In our submission to the committee, we've set out the waterfall approach that we take to age assurance, and, upfront, obviously, we have what we call a neutral age gate, so, when you try to create an account on TikTok, you get asked to input your date of birth. We don't try to probe anyone to put in any one particular date or another, or ask them what their age is. They have to, obviously, declare that upfront. If you do make it onto the platform, we have an ongoing monitoring system. It's a combination of what I'd call AI smarts and technology tools, human moderation and also reporting systems. If I can just step you through that very quickly in the least jargony way possible: on some of the technology tools that we use, it's obviously looking at someone's profile pictures and the type of engagement they have on platform, as well as, potentially, messages that they send. If someone, for example, sends a message to someone saying, 'Happy 15th birthday,' that would cause a flag in our system. So that's part of the technology suite that we utilise. Then, obviously, we have human moderators as well. As part of the implementation of these laws, we've invested quite significantly in additional training for our human moderators who are focused in particular on age assurance, to make sure that they are up to scratch with the new requirements in Australia and know what to look out for. Then, obviously, we have a suite of what we call reporting functions as well. We have a particular under-16 button in our app, so that, if you suspect that someone is underage, you can simply report it in through that, directly. But that is not just for people who use the app. We also do have a web portal, so that, if you are not on the application itself but you have suspected underage users—that there's someone that you think might be online that shouldn't be—you can report that in to us directly as well. Senator ANANDA-RAJAH: You might have mentioned this in your opening statement, but have you declared how many user accounts you've removed? I might have missed it. Ms Lofstedt : That's okay. It was in our submission. As we were leading into these laws coming into effect, we deactivated 550,000 accounts upfront. On top of that, we continue to take accounts off platform, if and when we find them. So we take proactive enforcement actions. There are approximately 24,000 accounts a month, since the implementation of the law, that we continue to take down. Senator ANANDA-RAJAH: Do you have a sense of how many users try to get back onto the platform? Ms Lofstedt : That's not a data point that I have with me today, but, again, I come back to this: it is an iterative process. Unfortunately, young people do try to make it back onto the platform, which is why we have this ongoing process of improvement and ongoing monitoring to make sure that we deploy the technology suite that we have available in combination with human moderation coupled with proactive reporting from parents, guardians, teachers and people who basically see suspected activity online that they can report in to us. We continue, then, to take accounts down as— Senator ANANDA-RAJAH: Do you have any data on reporting from adults on your platform? Ms Lofstedt : It's not a data point I have with me here today, but we would have some data on that, yes. Senator ANANDA-RAJAH: Okay. If you could, maybe supply that to us on notice. Is it an easy process for parents or teachers or people in authority to navigate? Ms Lofstedt : Yes, absolutely. As I said, in the app itself, you have a very simple way of reporting through our under-16 reporting button. Even if you're not on the platform, there's a web portal and a form that's very, very straightforward where you can go in and report to us. Again, we look at every single one of those through human eyes, and we basically try to bottom that out as quickly as possible. Senator ANANDA-RAJAH: My final question is to Snapchat. It was previously reported that Snapchat was using its facial age-estimation tool without a liveness test being turned on, meaning the tool was not primed to detect whether or not the face being shown was just a still image. Why wasn't it turned on, and what advice did you receive from your third-party age-assurance provider about the liveness test? Mr Au : I can confirm that our facial age-estimation tool does have liveness detection. It has always had a liveness detection, since the social media minimum-age law has come into effect. For our facial age-estimation, we use a third-party provider. It is a provider that was a part of the age-assurance technology trial. They were assessed at the highest technology readiness level. I am aware of the reporting, potentially in the media, around the liveness issue. My recollection is that we sought to correct that report immediately. Senator ANANDA-RAJAH: Okay. Thank you for that. CHAIR: Deputy Chair. Senator HANSON-YOUNG: I appreciate you all being here today—virtually. It would have been good to have some of you in the room. Firstly, could I ask Meta to clarify what the 750,000 accounts were directly related to. Ms Garlick : Under-16-year-old accounts that we've removed across Facebook and Instagram. Senator HANSON-YOUNG: You started off with 500,000, and you're saying that, since the social media ban first came into play, you've removed an extra 250,000 accounts. Am I understanding that properly? Ms Garlick : Yes, that's correct. Senator HANSON-YOUNG: And you don't have a figure of how many accounts you suspect are still underage? Ms Garlick : No. From our perspective, if we suspect it's underage, we're removing it. Senator HANSON-YOUNG: How many accounts are you removing per day? Ms Garlick : I'm not sure that we have a daily figure. We've received, I think, eight information notices. We're getting information notices monthly from the eSafety Commissioner. We have disclosed this new number of 750,000 based on these monthly information notices that we've been providing. Senator HANSON-YOUNG: So you've got the figures for at least each month. Ms Garlick : That's right. Senator HANSON-YOUNG: How many would you have removed in the last month? Ms Garlick : I don't have that to hand. I'm happy to come back to you if that's helpful information. Senator HANSON-YOUNG: Yes. We're here for a little while, so, if someone from your team could text that through to you, that'd be helpful. Could I ask the same of Google. So that we're clear, how many accounts overall have you removed since the inception of the ban? Ms Lord : Just to be clear, we have a single Google account that's used to access multiple services, and that includes YouTube, so what we've done in response to this law is block access to YouTube via that Google account for suspected under-16 users. What I can share is that, since 10 December, 740,000 accounts that had tried to use the platform since that day and that we suspected were under 16 have been prevented from signing into YouTube. Senator HANSON-YOUNG: That's pretty comparable to Meta, then, in terms of 740,000 to 750,000 overall. How many under-16-year-olds have Google accounts? Ms Lord : I don't have that data point to hand; I'm sorry. Senator HANSON-YOUNG: I'd be interested to know how many, seeing as it's a different scenario to, say, Meta or TikTok, in that regard. Ms Lord : It is a different scenario. Just to be clear—obviously, people own Google accounts for a whole host of reasons. That could be because you've created a Google account to set up Gmail or that your parents have created a Google account for you when they're setting up an Android device. There are multiple reasons, not just necessarily to use YouTube as a service. That's why we've given you that figure, which is a way more accurate representation of, really, the impact of this legislation. Senator HANSON-YOUNG: If you could take it on notice, I'd be really interested to know how many broad Google accounts exist for under-16s— Ms Lord : Understood, yes. Senator HANSON-YOUNG: for all of those reasons. Thank you. Do you have data that says how many people access YouTube logged out? You must have that data somewhere. Ms Lord : I don't have that data, and I don't believe I've seen that data either. Senator HANSON-YOUNG: So you don't keep a record of how many people use YouTube in the logged-out state? Ms Lord : Not that I'm aware of, but I'm happy to take that question on notice as well and see what we can provide. Senator HANSON-YOUNG: You must have it, because there must be a difference between how the advertising works for a locked-out account versus a logged-in account, surely. Ms Lord : Of course, because we apply different restrictions to advertising depending on the age of the user. As you'd know, we don't target ads to under-18s, for example. We have additional protections in place—or we did—for under-18 account holders in Australia with regard to advertising. I'm happy to take that on notice and see what we can share with you that might be useful. Senator HANSON-YOUNG: Okay. I'd like to know what level of revenue Meta, TikTok and Google all gather from gambling advertising. Ms Garlick : I'm happy to take that one on notice. I'm sorry; I don't have that to hand right now. We're happy to share what we can. Senator HANSON-YOUNG: Google, do you have the figure? Ms Lord : I'm in the same position. It's not an issue that I cover in my role, but we'll see what we can provide on notice. Senator HANSON-YOUNG: TikTok? Ms Lofstedt : I came prepared for the social media minimum age ban legislation, so that's not a data point I have with me today. Senator HANSON-YOUNG: I'd like it answered; it relates directly to age verification because of the next iteration of how age verification may or may not be used, given the government's planned gambling legislation. Ms Lofstedt : I'm happy to take it on notice. Senator HANSON-YOUNG: Snapchat, do you take any revenue from gambling advertising at all? Mr Au : There are gambling advertisers on Snapchat. Gambling ads are not served to users aged under 18. Senator HANSON-YOUNG: That's for those users you know are under 18. Mr Au : That's correct. Senator HANSON-YOUNG: How does Google determine whether somebody who is underage is receiving gambling advertising, whether that's on YouTube or they're searching on a website and Google ads are popping up? How does Google determine whether the advertising on gambling is age appropriate? Ms Lord : Across Google's products and services—and bear with me, because my area of responsibility is obviously YouTube—we rely, as we've talked about, on the age associated with the Google account. We take that to be a strong signal for a number of reasons, and we have protections in place to prevent users from changing that age date associated with the account, depending on the circumstances. At the Google account level, in January we rolled out age inference models across the Google account to detect whether a user—based on signals on that account—is over or under 18. Where someone has a declared age of over 18 or an age associated with that account of over 18 but their behaviour suggests that they are in fact under 18, we will default them into a minor experience and give them an opportunity to verify their age to correct that if that assessment is wrong. Senator HANSON-YOUNG: This is a question for Meta. Are young people being served up gambling ads across your various platforms? Ms Garlick : No, Senator. We do have additional protections in terms of advertising that is shown to people that we believe to be under 18, and we've spent many years building out what we call an adult classifier that looks at a range of different signals to identify if someone is over or under 18. I do think it's important to note in the context of our discussion of the social media age ban law one of the things that I think we have shared with earlier versions of the committee's inquiry as the law was taking effect, and that is that 16 is a globally novel age boundary. That's why we're all talking about these iterative approaches that we're taking as we're trying to build up new signals and markers around this globally novel age boundary. It's not quite a simple comparison of how the industry has worked towards the 18 age boundary versus this new one that we're working towards now. Senator HANSON-YOUNG: But if you're over 18, you can just see gambling ads on your various different platforms, can't you? Ms Garlick : You can see ads that comply with our policies and with applicable laws and advertising standards, yes, and there are various controls that people have if we're not serving them ads that are interesting. Senator HANSON-YOUNG: We've been given all the data and analysis to show that 80 per cent of young people under the age of 16 are still on your platforms. So who's failing? Whose fault is that? Meta's removed 750 accounts. Google's blocked 740,000 accounts—if you're under 16—from getting onto YouTube. You're telling us you're doing all this stuff, but 80 per cent of under-16-year-olds are still accessing your platforms. Whose fault is that? I'll start with Google. Ms Lord : I had a feeling you were going to ask me to go first, so I was reaching for the unmute button. There are a few things I'd like to share in response to that. Firstly, I think you might be referring to the eSafety Commissioner's latest research. That was clearly at a point in time, in March. I would say—as we've discussed here today—the technologies that we are using to detect users at the 16 age boundary are evolving because that is a novel and complex age gate. I can say that, on our end at least, we've been working towards improving those systems since 10 December, and we've been clear with the government about that from the get-go. I remember that conversation when this law was first mooted, before we even had the age of 16—that anything outside the adult-to-minor age boundary was new and that detecting a user at a distinct age of 16 was also new. We are iterating in that space and have been taking steps from day one and will continue to do so. We do keep the eSafety Commissioner updated on that work. Senator HANSON-YOUNG: Does Meta take any responsibility for 80 per cent of young people still accessing your platforms regardless of the fact that they're not meant to be? Ms Garlick : We are certainly doing everything we can to identify and remove and to have anticircumvention measures in place. We're continuing to invest and work to comply with the law. I do think, as well, sometimes it's challenging—I'm not sure which particular report you're referring to, but there can be differences in terms of people accessing versus having an account. I do think survey data is sometimes challenging in terms of 'stated' versus 'revealed'. So there could be a number of factors in place that are contributing to that. But certainly, from our perspective, we're taking our obligations seriously and working to continue to invest and make sure that we're doing everything we can to identify and remove accounts in compliance with the law. Senator HANSON-YOUNG: I've got one other question, and that is about whether you would all be prepared to give your users the option to opt out of the pushed and promoted posts and algorithms, as per people being able to have control over their own feeds. I'd like an answer from each of you. Ms Lord : I'm happy to jump in here. We appreciate the desire for our users to have control over their experience when they're on YouTube. We do already have tools in place to enable those users to shape recommendations on our platform, and that also includes the ability to turn off personalised recommendations on the home page and on the shorts feed. What I would say, though, is that adjacent to this conversation is a suggestion that we wouldn't agree with, and that is that algorithms are inherently harmful. I think, on our platform, algorithms are one of the tools or recommendations that we use to maintain a safe and responsible platform. They also allow us to serve content to users that's useful and relevant to them, given that YouTube is quite a different offering to other services—we don't connect users to content based on a social network. I think, when we have these conversations, they often centre around concerns about content, and our view is that, if the concern is around content, then we should be having the conversation about content rather than the feature itself. Senator HANSON-YOUNG: Does Meta think that individual users should have control over their feeds? Ms Garlick : Yes, absolutely. I think that, to build a little bit on what Rachel said, there are different categories of algorithms that are running across our platform at any given time. There are ranking algorithms for connected content, there are recommendation algorithms for unconnected content and then, obviously, we have a range of different safety and integrity algorithms. But, with respect to what people see in the feed, we survey tens of thousands of people every day across the site to understand whether they are or are not interested in the content that they see. There's the ability to reset your algorithm. On every post that you see, you can understand why you're seeing it. And then you can certainly pause having content recommended to you. There are a range of different controls there so that people are finding the content that they're seeing and engaging with useful and helpful. Senator HANSON-YOUNG: So why don't you just give people the option to opt in and out already? Ms Garlick : Well, there are a range of different controls that people can have to change how they're seeing. For example, on Instagram, if you only want to see content from your favourites, then you can control that. So there are different tools that people have to be able to have the control of what they see in their feed, if we're not getting it right in terms of how ranking or recommendation algorithms are working. Senator HANSON-YOUNG: But why can't we turn off the ads and the sponsored posts? Why can't a 17-year-old say, 'No, I don't want to be bombarded with advertising, so I'm going to turn it off.' Ms Garlick : So we're talking about advertising, not algorithms? Senator HANSON-YOUNG: It's the algorithms that drive that content to that person's feed. That is the only reason you've got those ads showing up in people's feeds. They're being driven by the algorithms. Ms Garlick : Maybe it's helpful if I just clarify that. Senator HANSON-YOUNG: No, I want to understand why it is that a young person is not able to opt out of your promoted posts or your algorithmic advertising. Ms Garlick : Sure. Maybe I can clarify. In terms of advertising, there are very limited opportunities to advertise to young people—or people under 18. As for the way in which that works, it's based on location and their age. So it's a much more limited offering. It's not the same experience as for an adult. And then, with teen accounts, there are even greater controls around the types of content that people can see so that people aren't being exposed to content that might not be appropriate for someone who is under the age of 18. So we do have a number of controls in place to try to address what I think is the concern that you're asking about. Senator HANSON-YOUNG: Why can't a 20-year-old opt out of being served up gambling ads? Ms Garlick : They can. If you're served an ad that you don't want to see in your feed, then you can opt out of seeing that. Senator HANSON-YOUNG: No; you can say, 'I don't want to see this anymore.' You can't say, 'I don't want to be bombarded with gambling advertising.' You'll just get served up the next gambling ad. Ms Garlick : I'm happy to share more with you about ad preferences. You can actually put in controls to try to prevent certain types or categories of advertisers from being shown to you, and I'm very happy to share some more details about how that works. I think it's a really important feature, and it's helpful for us to make sure that people are aware of it. Senator HANSON-YOUNG: So you're telling me that you actively allow people to say, 'I don't want to get advertising in my feed,' as an adult? Ms Garlick : No. I'm saying that you can say that you don't want a particular advertiser or category of ad. You can certainly have those kinds of controls. But it's an ad supported service. Senator HANSON-YOUNG: How far down the account is that hidden? How many clicks do you have to go through to get that option? Ms Garlick : Every time you see an ad, there's a drop-down button where you have the option to hide the ad, send feedback about the ad or hide ads from that advertiser and then click through to your ad preferences so that you can set them. I'm happy to share details with you about that. We've certainly tried to make it front and centre for people as they're engaging with content on the platform. Senator HANSON-YOUNG: I don't think 'front and centre' is an honest and genuine description, with all due respect. But the chair wants me to hand over the call. CHAIR: Thank you. Senator Henderson. Senator HENDERSON: Ms Garlick, if I could start with you, we've heard a lot of evidence, including today, about the online harms being caused by social media platforms, including those run by Meta, Facebook and Instagram. Do you concede that your platforms continue to cause harm online? Ms Garlick : From our perspective, we are trying to provide services that people find helpful and useful and want to continue to use. That's why we invest significantly in tools and protections to remove harmful content and to make sure, particularly for young people, that they are placed into these higher settings on teen accounts. Senator HENDERSON: Can I just go back to my question? Do you concede that your products are still causing harm online? Ms Garlick : That's certainly not the way we've designed them, and that's not the way that we have invested in the many different content and integrity measures. Senator HENDERSON: I would put to you that your social media feeds are absolutely designed to keep people addicted to the feed. It is absolutely disingenuous for you to say otherwise. The addictive nature of the social media feeds, including their being driven by the algorithm, is a key design feature of your platforms. Isn't that the case? Ms Garlick : Engagement is only a very small signal that goes into how the algorithm functions. I'm happy to share more details with you from the systems cards, which show the range of different things that factor into it. But, from our perspective, if people don't find our services useful and they don't feel good when they use our services, then we'd not be able to continue to provide them and have a business. And so we have every motivation to try to make sure that our services are useful and helpful to people. Senator HENDERSON: Well, that's very gracious of you to offer not to continue to provide your service, including, I guess, in Australia! But the bottom line is—and I refer to the decision that was just made in New Mexico, with $1.3 billion in damages awarded against Meta. I understand you're appealing this decision. But the New Mexico Attorney General has accused Meta of designing its products to addict young users and failing to protect children from sexual exploitation on its platforms. The judge in this case ordered Meta to impose a range of youth safety measures, including monthly limits on teens' use of Facebook and Instagram, restrictions on notifications, tighter controls on adult contact with minors, safeguards for AI chatbots, and enhanced review of child sexual abuse reports. Are you going to implement those measures, in accordance with the court's order? Ms Garlick : Obviously, we work to comply with applicable laws and with court decisions. I've mentioned a number of the features and controls that we have in place to protect young people, and this is something that we continue to invest in. Senator HENDERSON: I would put it you that we've heard evidence as recently as a couple of weeks ago in relation to this particular case that Meta is fundamentally failing in its responsibilities to keep children safe online, Ms Garlick. You can say, 'We're doing this and we're doing that,' but the evidence is that you're not. The fact that you haven't even made available the Apple API in relation to safety tools demonstrates that. What you've done has been done kicking and screaming, rather than being proactive to keep children safe online, and we've got further evidence of this. My understanding is that Meta has got hundreds of court cases against it currently. Is that correct? Ms Garlick : I don't have any figures to hand on that. Senator HENDERSON: That's what's been reported publicly. Before I go back to some of the particular concerns about the social media ban, I want to raise something that I raised with you in a meeting, and that is that Facebook is continuing to publish horrific scams on its platform—scams which are promoting the logo of the CBA, Commonwealth Bank of Australia, along with Tom Piotrowski. These scams are driving certain Australians to use a WhatsApp group, and collectively there has been a loss of millions of dollars. Now, how can you possibly publish financial information on your platform when you know that it is a scam? Ms Garlick : I'm happy to share an update with you about that particular issue. After you raised that with us when we met recently, we did contact the counterparts that we work with at both the CBA and NAB, because we looked into this scam—I think it was late last year—and disrupted it, and then it reappeared. We have now taken further action. So it was a really helpful insight from you to raise that concern. Part of the new scam prevention framework will include the actionable scam intelligence signal sharing. We've already initiated a lot of that with the banks, and we continue to work really closely with them to make sure that, as and when new scams emerge, we're taking action on them as fast as possible and doing everything we can to disrupt them at a much more strategic network layer. Senator HENDERSON: Ms Garlick, can I just make it very clear. Your actions are not working. These ads are continuing to run on Facebook. I saw one just a couple of weeks ago on that particular day. Why are your systems not able to detect these logos? This is a prominent Australian bank. I also condemn the Commonwealth Bank for not taking action when its logo is misappropriated, and as a result Australians are losing millions of dollars. Why are you not able to detect what is basically the work of criminal networks or syndicates, I assume, misappropriating the logos of Australian banks and promoting them on your social media feed? Ms Garlick : In relation to scams, we have a range of different systems that we use to try to prevent them from running. We've also got our financial services verification tools rolled out for ads, to give Australians greater confidence if they do see ads related to that. And then we also have the— Senator HENDERSON: Okay. So why— Ms Garlick : Sorry. Senator HENDERSON: Can I just stop you here. If you've got financial verification tools, how are you running an ad with the CBA logo before you've verified that it is in fact a CBA ad? That's what's going on, and that is what's driving millions and millions of dollars of losses for Australians. Many Australians have lost their life savings as a result of just this one scam. Ms Garlick : Senator, why don't I come back to you with the details about the actions that we've taken in relation to this particular scam. It's certainly something that we have been working very closely with our counterparts in the banking industry on. But, more broadly, I think we have discussed scams through various conversations as well, and I think this is something that does require whole-of-industry and whole-of-government collaboration, given the source and origin of these scams. It's something that we continue to get ahead of. Senator HENDERSON: No, no. Ms Garlick, I'm not going to accept that. You are publishing these ads. You have not verified that they are genuine. I also, as I say, condemned the CBA. You are continuing to publish these ads on your social media feeds. You are the publisher, you are directly responsible, and you are directly responsible for the losses that arise as a result because they are not genuine financial services ads. Something is going dreadfully wrong, and Meta is directly responsible for the harm that is being caused to Australians. I will welcome further information on notice. I need to be quick because I've got other quick questions that I want to ask you. We've heard that around 80 per cent of Australians under 16 are still active online. Meta and other social media platforms are required to take reasonable steps to prevent young Australians being on social media platforms, including yours. What is going so wrong? Have you worked that out? Ms Garlick : Sorry, was that to me? Senator HENDERSON: Yes, that's to you. Ms Garlick : Sure. We've been working to respect and comply with the law since it took effect. We were one of the first platforms that started notifying people that they would lose their accounts and providing them with information about how to save their contacts and download content. We started our enforcement on those accounts and finished it before the law took effect on 10 December 2025. We've been continuing to identify and remove accounts on an iterative basis and continue to improve and step up our proactive detection tools and our review of accounts. From our perspective, we're working to identify and action all known under-16 accounts from since before the law took effect, and we continue to do that each month. We've disclosed the number of accounts that we've taken action on to date. Senator HENDERSON: Sure, but the point I'm making is that some 80 per cent of teenagers are still on social media. So, the— Senator HANSON-YOUNG: Something's not working. Senator HENDERSON: What is not working? Why are so many under-16s still on social media? Ms Garlick : Unfortunately, I'm not sure I know the precise report that you're referring to, but, certainly, I think that there can be challenges in terms of some of the survey information that has been shared. From our perspective, when we're aware of someone being under 16, we're removing them. I'm not sure about how the survey is being conducted and whether there are differences between account holding or accessing, so it's hard for us to surmise that, but I do want to reassure the committee that, from our perspective, we are continuing to invest and are doing everything we can to comply with this law. Senator HENDERSON: I will have to move quickly to other companies. To TikTok, you referred to AI technology tools that you're using to identify young people on social media. How effective are these technology tools? What's the detection rate? What's the percentage of under 16 that you're actually detecting? Ms Lofstedt : It's important to understand here that it's not just any one signal at any one point in time. There's obviously a suite of signals that we look at here. As I laid out at the start of the committee hearing, we've got that age neutral gate. If you were to create an account on TikTok, that's basically where you start. We don't prompt people. They have to input their age and do so in an honest way. Unfortunately, from time to time, people are untruthful and they do make it onto the platform. That's why we have this technology suite of signals that we look at. So, again, it comes back to it not being just one signal in isolation. There are another number of them going through. Whether that's looking at account profiles—a profile picture, for example—or the types of accounts that any one person follows or the types of videos that anyone engages with in any one point in time, or if someone sends a message wishing someone a happy 15th birthday, for example, that's the type of signal that we would pick up. So it differs, and, as I said, there are a number of them working in parallel as well as— Senator HENDERSON: Could you please, on notice, provide us with that data? Your various AI or technology tools and the extent to which they're working or not working—if you could provide us with that data, please. Ms Lofstedt : Yes, I'm happy to take that on notice. I'm just mindful, as I said, that there are a number of signals that run in parallel, so it's not a stop-start approach. Senator HENDERSON: That's fine. We're happy to get all that data on notice, if we could, please. The eSafety Commissioner says seven out of 10 under-16s are still on social media and circumventing the ban. Do you know why there is such a large circumvention rate, from your perspective? Ms Lofstedt : From our perspective, we take our obligations under the law exceptionally seriously, and we've obviously put in place the waterfall approach as part of the recommendations of the guidance from the eSafety Commissioner. Senator HENDERSON: Do you know why so many young Australians are able to circumvent the social media ban? Ms Lofstedt : Again, I come back to the fact that, at TikTok, our approach is that, as soon as we are made aware that there is an account that is potentially underage, we investigate, we take enforcement action swiftly and we deactivate those accounts. Senator HENDERSON: This is across the board, this data, but it can't be the case that every one of the social media platforms is properly complying with their obligations when there's such a high circumvention rate. You must know why you are failing. Ms Lofstedt : I can't speak on behalf of other platforms, Senator. I can only speak on behalf of TikTok, and I know that the process we have in place is an ongoing and iterative process, and we have put that waterfall approach in place to make sure that, when we become aware that there's a potential underage user online, we investigate promptly and we take immediate enforcement action. Senator HENDERSON: I'll quickly go to Ms Lord. Ms Lord, I hope you understand—just to be clear and transparent, you are married to Senator Dave Sharma. I always like to put that on the record. Ms Lord, how many Google accounts were shut down or disabled in the lead-up to the social media ban coming into place? Ms Lord : I don't have that figure with me, and, to be honest, we didn't shut down Google accounts. We prevented those Google accounts being able to sign into YouTube. As I've shared, a more accurate data point would be the 740,000 accounts that had attempted to sign in or were signed in from 10 December that we suspected to be under 16. Senator HENDERSON: But people don't have a separate YouTube account. Ms Lord : There is a single Google account that's used to access multiple Google services, including YouTube. Senator HENDERSON: How are you able to prevent people from signing into YouTube? Ms Lord : We use a multilayered approach here, like everyone else, as was recommended by the eSafety Commissioner. It means that there is no single point of failure, but it also means that we're relying on different signals. We can only use signals that are available to us on our service. They are going to vary between services—and also technologies that have limitations. As we've been clear in our earlier testimony here, this 16 age gate is novel, and we've been working to expand those systems to use the various signals that are available to us to detect those users at that 16 age gate since 10 December. But that work is complex. We are trying to do that in a way where we're ensuring that we are detecting those users in a way that's proportionate, that's effective, that's safe and that respects user privacy as well. There is a lot of work going on behind the scenes. That work, as everyone has said, has been iterative, and it's been going on since 10 December. That is why, when we're talking about this 80 per cent data point, it is a point in time. It's very early on, and we'd caution against overreliance on that at this point in time. Senator HENDERSON: Every other platform has been very transparent about how many accounts were disabled, restricted or shut down. For Google, I again ask that question of you, Ms Lord. How many Google accounts in total, leading up to the social media ban, were restricted, shut down or closed, please? Ms Lord : I don't have that number with me, but what I do have is the number that I've shared, which is a much more accurate number because it actually goes to— Senator HENDERSON: No, I'm not interested in— Ms Lord : the impact of YouTube. It's people that were using the service, not just holders of a Google account, because, as you'd imagine— Senator HENDERSON: Just to take you back to my question—to be frank, I'm not interested in what you think is relevant. I'm asking for that figure and I'm pretty sure that someone pretty close at hand in Google knows that figure. So I am asking for that figure, and I would expect that that figure be provided. Ms Lord : We'll take that question on notice, but I just want to reiterate that we can give you the number of Google accounts. People in Australia hold Google accounts for all sorts of reasons unrelated to the use of YouTube. But we will find something for you. Senator HENDERSON: I appreciate that. The reason I'm asking is that the numbers we've been given do not add up to five million and the only platform that has not provided its numbers transparently is Google. Facebook, Snapchat, TikTok—they've all provided how many accounts were closed down or restricted. Google has not provided that data, and so I'm asking for that data because we need to work this out. I think the figures that the government has used are very rubbery. In my view, there are not five million accounts that have been closed down. I suggest that there are a very substantial number of Google accounts that were restricted or shut down, but of course they don't all relate to YouTube, because there's no separate account for anyone who wants to access YouTube. Ms Lord : Correct. As we discussed, they all would have had access to YouTube, but how many of them were actively using or attempting to use YouTube? That figure is 740,000. Senator HENDERSON: We're trying to work out the discrepancy in what's been provided publicly. The fact is that Google has refused to provide this number and I'm asking for this number. I'm pretty sure it's pretty well known by you or within your team. We take this seriously. We are seeking this information. We know you hold this information, Ms Lord, and we would ask that you provide it to the committee on notice. Ms Lord : Okay. I mean, we provided data that was requested of us to the eSafety Commissioner. I understand they've exercised PII over those data points, but I'm happy to take it on notice and see what we can share. Senator HENDERSON: Yes. You don't have the benefit of that PII, so I'm asking for Google to comply with its obligations to this committee and provide that data please. Ms Lord : Sure. We will take that on notice and see what we can share with the committee. CHAIR: Senator Payman. Senator PAYMAN:   Good afternoon, everyone. This question's been already asked of Meta, but can you please tell the committee how many documents relating to the social media ban your companies have shredded since this inquiry was established? Mr Au : We have not shredded any documents. Ms Lord : We're well aware of our ethical and legal obligations and we're not shredding documents. Ms Lofstedt : We, again, are well aware of ethical and legal obligations and are not shredding any documents. Senator PAYMAN: Thank you. Snapchat—Mr Au—in January, it was reported that 'underage Snapchat users are verifying their accounts by scanning the faces of people who are decades older and of a different gender'. What has Snap done to address this since January? Mr Au : I don't have a recollection of that particular report, but, as I previously mentioned, our facial age estimation tool is a tool provided by a third-party vendor. It is not developed in-house at Snap. It is provided by, again, a third-party vendor that was part of the age-assurance technology trial and was assessed as a part of that at the highest technology readiness level. That's the only information that I can provide on that. Senator PAYMAN: Can you take it on notice, because it's under the obligation when it comes to this ban? We'd love to know what's actually being done when there's that circumvention with people scanning other people's faces and being able to get access. It would be great if you could take that on notice. As drafted, the new information-gathering powers would allow the eSafety Commissioner to seek information from any person, including parents and even children. Is there concern about how broad these powers are in that regard, and is there a need for these powers to apply to parents or children? I'm happy for Meta to take the question and then we'll go to Google, TikTok and Snap. Ms Garlick : From our perspective, we're working to comply with the law. If we get requests for access to documents, then we'll work to comply with that. I don't think we have a particular position on the proposed amendments. I'm happy to throw it to Rachel. Ms Lord : From our perspective, the eSafety Commissioner already has significant investigatory powers under the OSA, and I want to be clear that Google has been cooperating in response to the seven notices we have received to date. For the committee's information, we have provided approximately 150 pages of information in response to over 200 questions in response to those seven notices. We received an additional notice yesterday. We do take these processes seriously, but we'll ultimately defer to you as policymakers on whether you think these amendments are necessary. Senator PAYMAN: TikTok? Ms Lofstedt : We're cooperating fully with the eSafety Commissioner. Our approach has always been one of cooperation and continuous improvement. As we set out in our submission under the provisions of the bill, that's really a judgement for parliament. Mr Au : In the interest of time, I don't have anything further to add to the views expressed by my fellow panellists. Senator PAYMAN: This is to Google. I'm interested in the part of YouTube's submission where it says: These long-standing age assurance systems had been trained to identify users under 13 drawing on audio patterns (such as voice tones) and visual patterns (such as physical size) in uploaded video content. I'm curious to understand this. How does YouTube tell the difference between a young person uploading video content of themselves and, say, a parent uploading a video of their child? Ms Lord : That's a great question. I'll have to give that some thought, to be honest. We have long had age-assurance systems operating to detect under-13 users based on the strongest signals that were available to us, and those are audio-visual signals in the videos themselves. To be clear, we've expanded those systems now after some investment to detect who we think to be 16-year-olds and under-16-year-olds. It would be a question of the interpretation of the law were it an account-level ban—is that account owned by someone who's under 16? We use that to detect the video content and then to make an assessment on who's owning the account. Senator PAYMAN: In another submission—I think it was from Project Rockit—they note the story of a young person who, now that they no longer have access to an account for YouTube, is being served an increased amount of content about self-harm and suicide. Why is it that banning young people from having accounts led to obviously undesirable stories like this? What safety features are lost through prohibiting young users from having an account? Ms Lord : This is something we've been on record on previously, and I don't think I need to rehash those earlier conversations, given where we are now. What I will share is that obviously this is an account-level ban. We respect the rights of users to access information without an account. We do provide logged-out access to YouTube, but that is a watch-only experience, and we block access to mature content. I find it surprising to hear that this user is encountering content that, quite frankly, shouldn't be on the platform because I would imagine that would be violative of our community guidelines in any case. It would also be age restricted and unavailable in a logged-out experience. Senator PAYMAN: What I'm trying to just understand is if you believe the decision of the government to not exclude YouTube has increased or decreased negative outcomes for young users. Ms Lord : We think that there are robust baseline protections in a logged-out experience, but we did have additional safeguards applied at the account level for under-18s. We also had additional tools for parent controls of those accounts, which have also been lost as a result of the social media ban. That included things like adjustments to our algorithm to prevent repetition of content that might be innocuous in a single view but harmful if viewed repeatedly by the 13 to 16 cohort, it included default wellness tools like 'take a break' and bedtime reminders and it included the ability for parents to turn Shorts off for their children. So those additional protections are gone for anyone under 16 and any parents who had set up parental controls on YouTube. Senator PAYMAN: So the onus really is on parents to be able to put those measures in place to protect their young ones. Ms Lord : Parents actually have lost parental-control tools on YouTube as a result of the social media ban, and that is because we had supervised accounts available, but that was account-level supervision and we can't offer accounts in Australia for under-16s. Senator PAYMAN: So there have been increased negative outcomes. Ms Lord : I wouldn't say it's increased negative outcomes. I'd say we'd lost the decades-long positive investments that we'd been making in age-appropriate safeguards for the 13 to 16 cohort and for parents supervising their children. Senator PAYMAN: Okay. Thank you, Ms Lord. For TikTok, I've just got one final question. In your submission, it states: Once an account is created on TikTok, our systems look at 'signals', to check for indicators that the account owner may not meet the minimum age requirement. We do this through a combination of Artificial Intelligence (AI) tools and extended human moderation capabilities. I'm curious to understand and know: in which country or countries are these human moderators based? Ms Lofstedt : It's a great question. As you know, we are a global platform and, obviously, we've got billions of videos quite literally every day uploaded on our platform. So, as I'm sure you can appreciate, our trust and safety team is stationed all around the world. We have what we call a 'follow the sun' model, to make sure that we do have immediate action taken on ground and we have that 24/7 safety approach. So the honest answer to your question is that it depends. We have people in a number of different jurisdictions, including here in Australia. Senator PAYMAN: Will you be able to provide the committee with a list of countries that you've got your moderators working out of. Ms Lofstedt : I'm happy to take that on notice. Senator PAYMAN: Thank you. CHAIR: We thank all of you for your evidence here today. You've taken a number of questions on notice. Please provide the answers to the secretariat by no later than 19 August. Thank you very much. .