KIRKHAM, Dr Reuben, Director, Free Speech Union of Australia PANE, Mr John, Chair, Electronic Frontiers Australia Inc. [by video link] SULSTON, Mr Tom, Head of Policy, Digital Rights Watch [by video link] CHAIR: Welcome. Thank you very much for being here. I understand that information on parliamentary privilege and the protection of witnesses giving evidence to Senate committees has been provided to you. We're running a little bit behind time and so, while I will invite each of you to make an opening statement, if you could, keep it as brief as possible. If you have something longer, please feel free to table it. I invite you to make short opening statements and, at the conclusion of those, I'll invite committee members to ask questions. Dr Kirkham : Thank you for inviting me to speak today on behalf of the Free Speech Union of Australia. We say that this bill is about an important issue—namely, transparency of how social media platforms operate. As it happens, we have strongly sought a lot more transparency, especially around online censorship. We think the public should have the right to know how content is both promoted and suppressed. X have announced this morning that they'll be providing some of this, so it is time for other social media companies to be required to join them. We are also concerned about the chilling effects of potential censorship, as well as the undermining of online anonymity. Without full transparency, we do not know if these age-verification systems actually respect privacy as they claim to do. We also think that the public should have a real right to choose which platforms they use. They can only genuinely choose if they are fully informed. Coincidentally, we're all winners. This is what is needed to protect people and to assess whether reasonable steps were taken to prevent under-16s using a platform. It feeds into any genuine consideration of a digital duty of care. How can we know what, if anything, is needed if we do not have a full picture as to how these platforms operate? We think that each platform should let all Australians see how content is being moderated, censored or manipulated. This should not be exclusively known to the eSafety Commissioner. All Australians have the right to know and the right to decide which platforms they use. We have some serious concerns about the bill as it's presently drafted in that it gives the eSafety Commissioner powers she should not have and, unfortunately, based on experience, are likely to be misused for alternative purposes. It is a regulator that you cannot trust whichever perspective you come from. This is a regulator that has appointed Professor Jeff Hancock, who previously did unethical psychological manipulation research for Facebook, as the person to oversee and run her research studies on the social media ban. We propose moving these enforcement powers to a different body that has the relevant expertise and is independent instead, such as the AEC. What we are proposing today genuinely advances online safety and supports free speech. It is something that everyone should be able to get behind, and we hope the Senate will be able to amend and improve this bill. Thank you. Mr Sulston : Senators, thank you for your work on this issue and for making time for us to talk today. Digital Rights Watch is an Australian charity that campaigns for freedom, fairness and fundamental rights for everyone who participates online. We recognise that there are many, many issues with the manner in which social media companies conduct their business, specifically around algorithmic-feed generation, privacy invasions to microtarget advertising and the proliferation of hate speech and mis- and disinformation. We've been campaigning for legislative and regulatory action on these and related issues for a decade. I'd like to recognise the important work that the eSafety Commission does in keeping Australian children safe online. All of that said, we remain opposed to the social media age ban for a number of reasons. It was rushed through parliament without sufficient public consultation and is half-baked experimental legislation. We have not seen any compelling evidence that it has made any positive difference to young people's wellbeing, but we have seen some evidence that it has caused harms to vulnerable children through increased call volumes to mental health helplines, as well as individual stories. We're concerned that it doesn't actually address any of the harms caused by social media companies and merely allows us to pretend that they're not happening to children. That's not due to insufficient compliance from platforms, insufficient regulatory powers or insufficient penalties. The social media ban doesn't work because it fundamentally misunderstands the manner in which the internet works. It treats censorship, of which age-gating is a type, as damage, and it routes around it. We need to regulate social media more cleverly than a blunt age ban. Our thoughts on this bill specifically are that we have not seen evidence that stronger enforcement of the ban will lead to better outcomes for Australian children, but we do strongly agree with the OAIC that age assurance is not a blank cheque to use personal or sensitive information. We cannot force social media platforms into increasingly invasive age-verification technologies in order to meet their minimum-age obligations. So we think that information-gathering powers for eSafety should be expanded to include the privacy commissioner, the OAIC, because privacy invasions for age verification are an issue, and we support the OAIC's submission to this committee in its entirety. Furthermore, we think that information-gathering powers should be limited to those people working for platforms and their suppliers, rather than private individuals, and that to go wider than that is a disproportionate overstep. In terms of penalties, we haven't seen any evidence of penalties having been issued or evidence that increasing them will increase compliance. We note, though, that the platforms have vast financial reserves, and we're not sure what difference a relatively small increase to the penalties will mean to them. We are all in favour of moving money from big tech corporations into the Treasury, so we don't object to this, but we would rather see a digital services tax than increased fines. I'd like to make two more comments. We support the Human Rights Commission's call for an early review of this legislation, rather than waiting for the current two-year review. I'd also like to support the Fix Our Feeds campaign, from whom I think you heard this morning, and their call for an opt-in to algorithmic feeds. That would reduce a huge amount of the harm caused by social media platforms and is a far better solution than the blunt instrument of the social media ban. I urge the committee and government to progress legislation in this area. Thank you. CHAIR: Thank you. Mr Pane : Thank you, senators and members of the committee, on behalf of Electronic Frontiers Australia. EFA is a not-for-profit digital rights advocacy run by volunteers, and it has been doing so since 1994. EFA opposes the proposed bill. This bill arises due to the existing blanket ban on under-16s accessing social media from 2024, which, with respect, has the appearance of a political syllogism based on a flawed problem definition and categorisation error. The 2024 bill, passed in an incredibly short timeframe, was essentially a blunt instrument born of populism and political expediency that has not worked in practice, as the EFA anticipated. Instead, it isolates and harms young people by removing them from their rights and agency. These harms are amplified in particular subpopulations, such as young people, people living in remote communities, LGBTIQ+, disabled youth and young people impacted by familial religious beliefs or other problems. EFA maintains its opposition to the federal government's policy framework on social media minimum age restrictions due to its reliance on an ineffective and harmful prohibition model. The existing framework seeks to solve vague and undefined digital risks by excluding users under 16 years of age from designated platforms, rather than addressing systemic design flaws inherent in commercial digital products and services that affect users of all ages, and not just children. To that point, the primary drivers of online risk are surveillance-based data extractive business models, engagement-maximising algorithms, dark patterns, harmful or extreme content, unregulated AI platforms and tools, and inadequate default privacy and safety protections affecting all users. An age based ban also requires continuous, population-wide age assurance, introducing significant privacy and cybersecurity risks for other users not subject to this regulation. In terms of rights impacts, the existing exclusionary framework denies youth their rights under recognised international treaties, drives them towards less regulated online environments, erects digital surveillance barriers for all citizens and cuts off essential support networks for vulnerable groups—specifically, LGBTIQ+, disabled children, and young people in households affected by domestic violence or substance dependency. Where we are right now, though, is what the big tech mantra of 'move fast and break things' has rendered upon us through our own regulatory action and inaction. In the context of social media as a digital product, it has not really resulted in a better world, nor in a more harmonious or safer world. Instead, we have a more unequal world, a more unstable world, a more surveilled world and an angrier world which, through deliberate technology design choices, has resulted in harm to both children and adults alike. In less than 20 years, big tech and social media platforms have shaped a world that increases political and interpersonal polarisation. It has been architected for social disharmony as a profit generator. It is a place where power is concentrated, where regulatory capture is a readily achievable enabler and where platform accountability is absent. There is a better regulatory alternative. EFA strongly recommends repealing age based exclusions in favour of a robust statutory digital duty of care, with a focus on the best interests of the user—strict mandatory privacy and safety-by-default laws, restrictions on engagement driven recommender systems and prohibition of behavioural profiles, accompanied by strong ex ante AI safety rules, not voluntary guardrails. These activities will not be a panacea. They must also be enhanced through continuous digital civics education through primary and secondary schools and more educational resources made available to parents and carers. In closing, to solve our current society-wide problems in social media and Big Tech, our lawmakers must pass proper, fit-for-purpose laws. To do that, we must first dispel the commonly accepted myths of technology, which have resulted in an absence of regulation or regulation that's not fit for purpose. Those myths include the myth of technological exceptionalism, the myth that regulation stifles innovation, the myth of technological neutrality and the myth of the privacy paradox. Ultimately, true digital safety for everyone is achieved when we strip the platforms and big tech of their power to exploit the user rather than stripping the user of their power to navigate and participate in the digital world. CHAIR: Senator Henderson. Senator HENDERSON: Dr Kerkham, thank you so much for your time today and for your submission. We've got quite limited time, so I'm going to cut to the chase on what is a pretty detailed submission that you've made. You've made a number of recommendations in relation to transparency, including the publication of social media platforms' algorithms. Can you summarise why that is so important? Dr Kirkham : How would you regulate these platforms if you don't know what they're doing? How would people choose which platform to go on? You fundamentally need to know what they are doing to understand what a reasonable step is. To be able to enforce this legislation, one of the matters you need to know is what these companies are doing to their users. If it's evil or particularly abhorrent, then the standards that might be expected to keep children off them would be higher. If it's the other way around—if it's a benign platform—then the standards are the other way around, but you won't know that until you understand how they're manipulating people. People should have the right to know. If I'm being psychologically manipulated or I'm only being shown certain content through censorship or otherwise, why should I not get to know this? As an Australian, it should not be being kept secret from anyone. If you want a digital duty of care—I'm not a supporter necessarily of a digital duty of care—and if you want to protect the public, the first step is to make them informed. That doesn't even require a duty of care; it requires a duty of transparency. Senator HENDERSON: If there was the intention to psychologically harm children or others on a particular social media platform, do you think that would be evident in the details of the algorithm? Dr Kirkham : Yes. You would be able to at least see what it's doing. You'd at least be able to have the chance of proving it. If you were a victim of that, potentially you'd be able to bring a claim. You probably need more to make it more obvious, in that the social media companies are doing what's called A and B experiments all the time on users. Facebook's notorious for that, including Jeff Hancock's emotional contagion study. They know how they're manipulating their users. Require them to publish the data. Require them to publish what they are doing to people. That would be basic scientific research ethics. If it were done by a university, you would actually inform the people who participate in your study, even if they were unwitting users in this case, what the results were and what the implications are for them, rather than hiding it in an office somewhere or on a company server. Senator HENDERSON: It is the algorithm which is helping to drive that harm in your view? Dr Kirkham : There are features in that algorithm that are helping to drive that harm. There are also features in devices. For example, if you're protecting children, one of the most obvious ways to protect children is to stop them having smartphones. If they were just accessing the internet on a laptop instead, parents could supervise them, but the addictive quality is also reduced because you haven't got this device notifying you on a screen that's very confidential and that you can't be passively supervised on. There are a mix of measures that could be taken, but I'm not convinced, for example, that a social media ban was the best step. Making the social media platforms a lot more transparent would be the way forward, as would stopping kids having smartphones. Senator HENDERSON: You're also advocating for social media platforms to publish their policies, practices, algorithms including model weights for relevant AI systems—I would like you to explain what that means—and source codes. Why are those— Dr Kirkham : Why those particular elements? Senator HENDERSON: things so important? Dr Kirkham : If you wanted to actually check what they were doing, you need the source code. The algorithm might not be fully implemented by the source code. If you want to run an experiment to actually see how effective a particular intervention is or to understand their effects, you won't have millions of dollars to go and build that code yourself necessarily; therefore, to actually subject them to scrutiny, you really need those materials. Similarly, if you want to see if an age-verification algorithm works, most of these are trained models. How would you run an experiment on them without actually having the trained model in question? How would you see, for example, if a system worked for Indigenous people? There's hidden trial data in the age-assurance technology trial that actually showed it didn't work at all for Indigenous people, at least most Indigenous people. Therefore, you've got these tools that are being recommended, but they don't work, and that makes the ban unenforceable because it's not a reasonable step to go and implement these tools. That is why the social media ban cannot be enforced at all, because tech companies can defensibly say, 'We've done the reasonable steps.' Senator HENDERSON: I hear your concerns about the social media ban, but should the underlying law and the requirements of compliance on digital platforms be tougher? Dr Kirkham : Absolutely. In terms of their transparency, it's a winner all around. Free speech advocates want the platforms to be more transparent. Parents want the platforms to be more transparent. Everyone wants the platforms to be more transparent. And it doesn't involve things like jawboning. The current problem with the legislation is that the eSafety Commissioner selectively uses her powers. Will she be as aggressive with Meta as with X? Probably not, given that she employs one of Meta's flunkies to run the social media study, Professor Jeff Hancock. If every platform is required to publish it to the public, then it cannot be selectively enforced, everyone can see in real time, and everyone can make decisions. Parents can immediately, if they want to, advocate. They can see the data, and they can advocate. They don't have to wait. Senator HENDERSON: Do you think, if Meta, for instance, and others were required to publish their algorithms and their source code, that that would be a real constraint on some of their insidious practices which, as we know, are harming children? Dr Kirkham : Absolutely. There's social pressure, but also it makes it far easier for someone who's been harmed to litigate that case because they've got the data and they've got the evidence. It's like consumer law. You wouldn't let all these secret things happen with your food. You wouldn't let this happen for any other product that you're using, so why are they not required to be honest about how they're censoring and manipulating content for engagement and for other reasons as well? Let's just have it all out there. Senator HENDERSON: And, just very briefly, you also want to see transparency in relation to what words and what phrases are censored by the various platforms. Why is that so important? Dr Kirkham : Yes. Otherwise, you'll end up with censorship being hidden in the system. The European Union has a legislative transparency report, but in practice it doesn't tell you how content is being skewed. That could be people with gender critical views. It could be people with views on the Israel-Palestine conflict. Take your pick. We want to know that these platforms are being even-handed in what they're moderating and how they're moderating it. Also, for the social media ban, as it's been termed, to assess whether a step is reasonable, you have to actually assess how it's being moderated. A platform that, for example, is heavily moderated on certain topics is quite different to a platform that is not moderated at all. Senator HENDERSON: Do you believe that these changes to the legislation would help the eSafety Commissioner enforce the law as it currently stands? Dr Kirkham : Yes. Ideally, the further proposal that we made is actually to move this to the ACCC because that's a far more effective regulator. It's not politicised, and it's got experience and expertise of this type of litigation, which the eSafety Commissioner lacks. Senator HENDERSON: But, putting aside the regulator, you believe that this transparency would make it easier to enforce the current prohibition on under-16s accessing social media? Dr Kirkham : Absolutely. If this is out there, every academic and every researcher can evaluate the material. At the minute, if you've got a small team sitting in an office doing it quietly with no public pressure, it's not going to be that effective. In the eSafety Commissioner's office, I doubt they have anyone with expertise, let alone the capacity, to analyse all that information. Senator HENDERSON: Thank you very much, Dr Kirkham. I appreciate your evidence. CHAIR: Senator Payman. Senator PAYMAN: Good afternoon, all. My first question is to Dr Kirkham. You suggest in your submission that the age-assurance technology trial report contains citations that appear to be AI hallucinated. Last year, Deloitte partially refunded the Commonwealth for a report that contained false, AI-imagined citations. That was for a $400,000 report. The age-assurance technology trial report cost $6.5 million. Should the Commonwealth be seeking an explanation and/or refund? Dr Kirkham : Absolutely. This was an appallingly conducted trial. I'm a computer scientist by training. I used to be a computer science academic. Any academic who produced this work should be dismissed. It involves AI hallucinated citations. It involves a complete lack of methodological rigour. It involves hiding some of the results. I managed to take the trial data and understand exactly what was going on, and they were literally hiding the fact in their final report that they were promoting systems that let over 40 per cent of ten-year-olds through an age gate and the systems discriminate. Some of these are rating Indigenous children as being in their 50s and 60s—literally. A lot of the material in the report is pseudoscience. I would be very interested to hear from the people who authored that report what the actual citations were, if they were not AI hallucinated. Senator PAYMAN: Thank you, Dr Kirkham. Mr Pane, in your submission, immediately after stating your opposition to the increased penalties, you're at pains to note: This is not because EFA are friends of big social media platforms or BigTech in general. Do you feel that there has been an attempt by supporters of the ban to create a dichotomy, with parents on one side and big tech on the other, with very little room for nuance? Mr Pane : I think, definitely, big tech have been trying to control the narrative around this as much as possible, but I think they would be reasonably happy with the outcome of having to comply with an obligation of taking reasonable steps to authenticate the age of the users, because it's such a vague term. It's also a vague term that's been used in privacy law for over 25 years and has been a matter of some discussion and debate. Senator PAYMAN: This is a question to anyone on the panel. In January, the government was celebrating the 4.7 million accounts banned. In March, they discovered serious concerns with compliance by social media companies. Was there a big change in compliance in the two-month period, or was this an attempt to set the narrative and, when that failed, to find someone to blame? Mr Pane : Could I possibly take this? Senator PAYMAN: Sure. Mr Pane : I was a member of the age-assurance advisory board for the project. In fact, I was the only privacy specialist on that board. I raised some very strong concerns from the outset about the structure of the project, given that it was being run by an organisation which was the industry group for age-verification providers benchmarking against a standard that they developed without [inaudible] in my view, after spending some time on the board, a proper understanding of the requirements of Australian privacy law. It had, to me at least, in my opinion—and I've spent 30-odd years practising in privacy law—the attributes of a privacy-blocking exercise. It was remarked that the terms of reference of the project would exclude specifically how children might circumvent the controls that might be being used. Had that been done, I think we would have reached a completely different outcome. Senator PAYMAN: Would anyone else like to make a contribution? Mr Sulston. Mr Sulston : I have a very short point. The ultimate arbiter of whether the social media ban is successful or not is, first of all, the wellbeing of Australian children. Specifically, I think we need to look closer at vulnerable children—those who will be less well served by being banned from social media because they may depend on it to reach their people. That might be ethnic minority groups, LGBT youth or those who are living in remote locations. So counting how many accounts have been banned is not a useful metric in terms of measuring that efficacy. We don't know how many accounts there were in the first place, we don't know how many were held by children, we don't know how many were banned in error and we don't know how many of those children then immediately went and made new accounts on the same or different platforms. I note from the report that eSafety released a week and a bit ago on a Friday afternoon that market penetration for social media amongst children has decreased, but it's from something like 84 to something like 81 per cent. So clearly it has not made a significant impact on children's use of these systems. Senator PAYMAN: That's a very important point. Dr Kirkham, would you like to make a contribution? Dr Kirkham : Let's put it bluntly. I have no idea why they're not—show me the receipts for these 4.7 million accounts. I've not seen the receipts. I'm not convinced. If all these accounts have disappeared, why is the same number of kids still online? Why is there no statistically significant evidence as to kids moving off social media platforms? Any reduction would be a study design flaw. The eSafety study, for example, is asking the parents if they know their kid's on social media. If they're not supposed to be on social media, are as many kids going to be telling their parents? Of course not. Senator PAYMAN: I'll put the rest of my questions on notice. Thank you. CHAIR: On that note, we say thank you very much to the witnesses for your evidence here today. The committee appreciates the work that goes into preparing for these hearings. If you've taken anything on notice, please provide the answers by 19 August..