FLEW, Professor Terry, Co-Director, Centre for AI, Trust and Governance, University of Sydney CHAIR: Welcome. I understand that information on parliamentary privilege and the protection of witnesses giving evidence to this inquiry has been provided to you. I now invite you to make some opening remarks, after which we will get into questions. Prof. Flew : I'd like to begin by thanking the committee for inviting me to be here today and by acknowledging that I'm on the ancestral lands of the Gadigal people. I'm professor of digital communications and culture at the University of Sydney, and co-director of the Centre for AI, Trust and Governance, established through the Faculty of Arts and Social Sciences at the University of Sydney in 2025. My centre co-director is Professor Kimberlee Weatherall from the Sydney Law School. I'm also an Australian Research Council Laureate Fellow with a five-year project on mediated trust ideas, interests, institutions and futures. Our centre is dedicated to ensuring that artificial intelligence is developed, deployed and governed in ways that advance public value, strengthen trust, and support Australia's economic, social and democratic priorities. The centre is inherently interdisciplinary. We work closely with computer scientists and engineers, but our focus is more on the social, ethical, cultural and economic dimensions of AI. The ultimate purpose is to help shape an AI-enabled future in which technological progress is matched by strong governance, public trust and meaningful societal benefit. It's sometimes said that the time to act on artificial intelligence was yesterday. I prepare a lecture on this topic for students at the University of Sydney each year. In the slides that I prepared for those students in 2023, it was estimated that by 2031, artificial intelligence worldwide would be worth US$1 trillion. When I revised those slides in 2025, it was estimated that the artificial intelligence market would be USD$3.3 trillion by 2027. I'll go back and look at what that figure is next year, but it could well be significantly higher again. My colleague Dr Rob Nicholls and I prepared the submission that you have. Recognising that there are other inquiries taking place into such matters AI infrastructure and data centres, I want to focus my comments today on some particular issues. Those issues will be trust, what I refer to as the AI dividend, copyright and IP futures, and the distribution of opportunities and benefits. In his address at the University of Sydney outlining the federal government's AI strategy, which I had the privilege of attending, Prime Minister Anthony Albanese focused on three issues in particular: copyright, data centres, and the future of jobs and work. There are also related issues under consideration and which frame the terms of reference for this inquiry, including national security, AI sovereignty, critical infrastructure, the rate of AI adoption and associated questions of risk, and the capacity of middle powers, such as Australia, to frame what might be referred to as the third way in AI governance. Trust is a cross-cutting, multifaceted concern that runs through all of the terms of reference and should not be treated simply as a function of data security. Important as that is, Australian AI and data policy has sometimes paid insufficient attention to questions of fairness and the distribution of benefits from new technologies. This runs the risk of eroding the societal trust on which technological adoption and, ultimately, our democratic system of government depend. Australia faces a significant trust paradox with regards to AI. In terms of attitudes to AI in general, Australia has been found in a few studies now to sit in the bottom third of countries surveyed. In the most recent study undertaken through the University of Melbourne's business school, only 30 per cent expressed trust in AI in general and only 17 per cent expressed high acceptance of the use of AI. At the same time, a recent survey by Anthropic on use of its Claude AI system—I acknowledge it's not an independent source necessarily—identified that Australians were proportionately among the highest users of the Claude AI system in the world. It was 6.3 times what would be the expected number of users. That is above Singapore, which is often taken to be an international benchmark in this space, and well above the United States of America. Australia accounted for 2.5 per cent of all Claude users, as compared to 0.5 per cent of global internet users and 0.3 per cent of the world's population. As such, Australia faces a significant AI trust paradox where high levels of personal use of AI systems—you can track this, for instance, by the number of magazines you can now buy in newsagents showing you how you can use ChatGPT to make money, organise your life et cetera et cetera—coexist with significant concerns about AI as a general-purpose technology. Australia also sits within the Asia-Pacific region, where rates of trust in AI and enthusiasm about the use of AI are among the highest in the world. In our submission, we propose that the Office of AI should work with independent researchers in the development of a global AI and digital trust index. Such an index, we would argue, is needed in order to reliably and independently identify changes in Australians' attitude to trust in AI and other digital technologies over time and to benchmark these trends against comparable international evidence, and the sort of framework I have in mind would be something along the lines of the highly respected Scanlon Index of Social Cohesion. That has been a resource for Australian policy-makers and the wider community since 2007. Whether Australia experiences rising or declining levels of trust in AI and, indeed, trust in social institutions more generally will critically depend on the equitable distribution of benefits of AI, or what we might term the AI dividend. There is considerable opportunity, I would argue, for local communities to extract benefits from investment in AI infrastructure as part of the social licence that is expected from large international companies operating on Australian shores. Investment in jobs in communities and in culture will be particularly important. In this regard, we particularly want to commend the importance of investing in First Nations communities and in learning from the insights that First Nations communities have generated over millennia about caring for country, and any lasting approach in the copyright space must also take account of the unique nature of Indigenous cultural and intellectual property and will need to be aligned with the Declaration on the Rights of Indigenous Peoples, UNDRIP, to which Australia is a signatory. I particularly commend article 31 of UNDRIP to this committee. With regard to copyright, a new copyright and intellectual bargain is urgently required for the AI era, but it is difficult because there has unquestionably been theft of the work of creative people, not to mention academics, by companies that have put their work into large learning models without permission or recompense. At the same time, the Copyright Act 1968 has long sat uncomfortably with the affordances of digital technologies and is particularly challenging to apply in the AI era. I refer here to the Australian Law Reform Commission's review of copyright in 2014 for background reference. One of the reasons that AI presents particular issues is around the provenance of content, or the extent to which it can be clearly traced back to an original source, which becomes more and more challenging. We see this, for instance, in the decline in the use of links from search engines as a source of traffic to news sites, as news is increasingly sourced from AI systems themselves in the form of AI generated summaries of available journalism. The information is not coming from nowhere, but it is difficult to trace it back to its original source. This points to the need for new forms of licensing and the empowerment of creatives to have bargaining rights in dealing with tech companies. In a longer discussion, I could also talk about cultural policy and the need to think more about distribution as distinct from funding activity in its own right. One of the things we're not short of in the current era is content. There is also a leadership role for government in establishing protocols for the proper acknowledgement and recompense of creatives for the use of their work, and there are emerging models starting to appear which I could discuss. Ensuring that there's not only widespread AI adoption but active engagement with AI among the Australian populace will be critically linked to the distribution of opportunities and benefits. This is partly about skilling up the populace, not just in technical skills but also in understanding the ethical, moral and epistemic challenges associated with these transformative technologies. Skills development for AI needs to incorporate AI literacies and sovereign evaluation capabilities as well as skills and capabilities in AI safety, trust and ethics. A very interesting model is provided internationally in the South Korean AI for All initiative. AI for All provides an example of an equitable and inclusive national AI strategy which distributes the benefits of the new technology equitably and is also cognisant of sovereign capability. Put at its simplest, the South Korean AI for All strategy provides each Korean citizen with their own AI agent, drawn upon locally sourced large language models. There's more to it than that, but that provides the core of the approach. The public sector could also provide an important test bed for strategies around what drives durable adoption and equitably distributed productivity benefits. This could include explicability of processes and outcomes and a focus on data quality, interoperability and workforce capability. There are opportunities for public sector agencies to act as test beds for safer AI, with transparent risk assessment processes, data governance and assessment in independent research access, including the development of public sector LLMs. The final thing I'd like to refer to is around AI and democracy. There are real risks that AI-generated misinformation, as well as what has come to be known as 'AI slop', will further erode public trust in institutions and degrade the wider information ecosystem. There is no single agency that can address all misinformation issues and the challenges of what are being termed epistemic risks. One recommendation we would make is that the current political advertising laws be changed to prohibit the use of AI-generated deepfakes in election campaigns. Such activities go beyond the legitimate bounds of public campaigning and prove to be genuinely threatening to the individuals and groups who are the targets of them. The future of democracy is not only challenged by AI-generated information. There is also the question about whether a wider commitment to truth may be in decline, which can corrupt the epistemic bases of democratic societies. Greater regulation of AI companies is clearly important to addressing such concerns, but the remediation of communication through AI presents new challenges to our educational institutions and to knowledge media and communications institutions as well as those of government. It's vitally important, then, to address such challenges as a matter of urgency so we can have a trustworthy and vibrant society in an increasingly AI-enabled economy. CHAIR: Excellent. Thank you very much. There's quite a lot in that opening statement, and obviously the remit of the work you're doing in your research space is enormous—but really important. I want to touch on a couple of things—in particular, trust and, as you said, the trust paradox in that Australians are quick and early adopters of this new technology but also have some significant concerns about it. I'm wondering what you could provide the committee by way of recommendations to government around where the gaps are that we need to fill to help build up that trust. It strikes me that we will not be able to achieve the full benefits of AI—to fully grasp it—even if we have all the technical capacity and capability and a whole range of other things, if people don't trust it and people then properly use it. Would that be— Prof. Flew : I think that's right. I think there are three dimensions to it. I think there's the question of trust in the companies involved in the development and deployment of AI—the question of social licence. I think that requires regulatory responses, but not just regulatory responses. I think there are opportunities for engagement, particularly at the local levels of government, around some of these questions—around the conditions under which, for instance, data centres are built. Secondly, there's the point that AI incorporates a lot of different things. It incorporates everything from robots in the workplace to autonomous systems through to generative AI, which I think a lot of our commentary is about at present, and agentic AI. I can illustrate some of the issues in Australia around robots in the workplace. It was pointed out to me when I previously worked in Brisbane that Queensland is now the motor-vehicle-manufacturing capital of Australia—a point that's not widely known. Robots are being routinely incorporated into those production activities. Governments choose not to publicise that because the perception is that robots are put into these production processes at the expense of people. Even if the evidence doesn't point to that, the perception is very strong. Someone might, say, trust their own ChatGPT system but have concerns about autonomous vehicles—or driverless cars, as they're better known. The third thing is that these systems themselves have elements to them that can be quite persuasive to the user, I think it would be fair to say. They can be constructed in such a way as to flatter the intelligence of the user. This may be starting to emerge where people are perceiving AI generated information to be more trustworthy than humanly sourced information. This is coming up in some journalism studies. If you get the news from the ABC or the Australian, some will say, 'Well, they've got biases.' But, if it comes from a chatbot, it's perceived to not be humanly sourced. The ways in which prompting and so on work can generate a deceptive level of trust. So the challenge will be to maintain the value of the human while recognising the possibilities that are provided through AI enablement. I think that's happening anyway, and I think there'll be an iterative process around persuasion, if you like, of the wider social benefits as well as the personal benefits to be derived from these technologies. CHAIR: It's a really interesting observation. We hear often in this debate about the concern around a lack of human involvement or a lack of human control and, understandably, that it is a really concerning element of this technology. But, on the flip side, the conversations that people are having individually with their specific chatbots—they feel like they can trust those more as sources than they do a human source. That's a really interesting juxtaposition that is worth keeping front of mind. I am not necessarily sure of how we best tackle that, but it is something to be alive to. Prof. Flew : I think it's happening lots in large organisations. CHAIR: Yes. Jumping back to my question about what government can do, what is the role of government here, at the big-picture level? How do we help Australia play a leading role in shaping the governance not just here but globally? Does it lend itself to our at least having a say and being able to articulate to the Australian people reassurances of safety and the remaining human capability of the technology? What do we do in order to make sure that happens? Do you think that is something that is important? Prof. Flew : I do think it's very important. There's a very important role for middle powers and, very possibly, for developing collaborative networks among middle powers in this space. A degree of local sourcing of information is going to come to be important. In that respect, the copyright and IP issues come back because the alternative is not going to be the absence of large language models; the alternative is going to be the absence of Australian content in large language models. Some of the perceptions around AI refusal, if you like, are probably missing an important point. At the same time, there's clearly a need to develop mechanisms whereby the provenance of content can be better sourced and can be more transparently understood, so knowing not only where your information is coming from but what are the uses of your own information. Personally, I found out via the Atlantic story about the Meta Llama 4 data hack that 78 of my works were in the Meta Llama 4 large language model. Had I been consulted on that, I may not have objected. I have, personally, a different view as an academic on published papers, which I do see as public product, and sole authored books, where I do derive copyright income. So I think those questions around the capacity to legitimately opt in or opt out of these large language models are vitally important and have to be there. But I think that, for that to be there, there will need to be some evidence of a dividend or a return to creators along the lines of what the Copyright Agency Ltd has done very effectively in Australia now for many years. CHAIR: Yes, obviously that has been a pretty significant issue that we're tackling around how do we maintain or, quite frankly, enhance the control for creatives. To the points that you've been making around journalism, the ABC made some similar points about not linking back to the original source and what that means for journalism. But the extent to which we can use this opportunity to get a better outcome in that space whilst making sure we're protecting and not watering it down but actually—how is it that we are still in this frame where legitimate issues are raised around use of data and that sense of input versus output, in particular, but input for training. How is it that we can provide greater control for creatives, and what could that potentially look like? What's your sense there? Prof. Flew : I think there's a technical side to it that I'm not the expert on, which is about a kind of tagging of inputs into such systems. But I think that a degree of transparency can be demanded. I don't think it should be left as a matter of choice. I think that, where there is a web source for a particular piece of information, that should be accessible from the AI summary in the same way as when we use the Wikipedia site we're provided with quite extensive footnotes as to the sourcing of the information. I also think that licensing arrangements need to recognise legitimate uses of output, and I do note that the current copyright framework presents a variety of problems in that respect. Could I provide an example? CHAIR: Sure. Prof. Flew : I will have a paper out shortly that will be talking about deepfakes. In order to demonstrate that deepfakes are not something that was simply created with generative AI, I use a shot from the 2015 film San Andreas, or I would have used a shot from the 2015 film San Andreas, where Dwayne Johnson gets the speedboat past the Golden Gate Bridge before the cargo ship knocks over the bridge and destroys the city of San Francisco. There's a lot going on here that we would now refer to as deepfake; I think you'd find that the history of Hollywood cinema has been based around that. But, in attempting to use that particular image, I waited six months for advice from Warner Bros on whether a still image could be used in an academic book, and the answer was no. So I see no public interest being served there at all, nor do I see a case for the returns to the original producers, which have certainly been valorised 10 years on from a film like that. CHAIR: On deepfakes or what is happening now around generative AI, the use of AI and the level of transparency, there has been some thinking around a requirement to label when things have been generated by AI. What's your sense or reflection on that? Prof. Flew : I think it's desirable. Public concern around it probably depends on the image itself. As I said, with cinema, we're now used to, after well over a century, watching things that we know not to be real. As with other areas around misinformation, I think it's difficult to set a blanket rule. I think there's a need to identify a context in which it's important, and it seems to me that the context of federal elections is a particularly important one. Australia not only is a stable democracy but is highly respected internationally for the conduct of its elections. And I think it's very important that that is protected. I think that a prohibition on AI generated deepfakes in election campaigns is a valuable, proactive step that could be taken now. In fact, it very possibly should have been taken three years ago, but, as with many things with AI, there's a bit of catch-up. I think that acting now is very important because the capacity to generate what I've referred to as AI slop is now so much greater and so much more widespread. CHAIR: Thank you. Senator Pocock. Senator DAVID POCOCK: Thank you for your very thoughtful submission on your work in this space and for appearing. In the context of the recent OpenAI hack, I'm keen to ask you about accountability. How important is accountability in rebuilding trust? How important is actually having a system where there is accountability built in? It seems to me that when you have accountability, people make different decisions. Prof. Flew : Yes. Senator DAVID POCOCK: Looking at what's happened, OpenAI has now apologised and it seems everything's fine. It seems there are two standards. If you're a big frontier lab, you apologise. If you're an Aussie, you probably face the courts and have the book thrown at you. Should we be thinking about a framework that apportions responsibility and serves as a disincentive to this sort of behaviour so that, if it happens, somebody is accountable—so you can't just say, 'Well, it was a bot, and we didn't intend it'? Prof. Flew : Sure. You certainly can't have governance by public apology. We've seen that in the history of social media. The point would be—first of all, there's the tracking of this. My understanding is it took quite a while at all stages of the chain to identify that anything had happened. Clearly, the security systems domestically need to be upgraded to capture this. I think also that companies such as OpenAI could apply conditions to their users on the nature of the uses that they make of their agents. And that would provide some legal framework in which you can hold a company liable for the behaviour of an agent, because—you're right—otherwise the chain of responsibility is just very, very hard to track. 'The bot did it' is not really a satisfactory answer. So terms of service could become something that's quite important in this regard—not that it would prevent all such instances, because these are very decentralised technologies, but it would create a chain of liability. We have seen terms of service have impacts with other digital technology forms. I think, for instance, about the changing relationship that YouTube has had around content over a 20-year period. Senator DAVID POCOCK: You talked about AI agents—autonomous agents, agentic AI. How should we be thinking through some of the very concerning reports of people's interactions with chatbots, where they're talking to it. We've heard stories of self-harm, of it directing a person—often a young vulnerable person—to potentially carry out an attack on someone. In one instance, I read about how it told a user how to commit suicide. Where's the responsibility there? Should the company be responsible for what ChatGPT or Claude is telling them? How should we be thinking about that as a parliament? Prof. Flew : I think it's going to be a framework that will be adaptive over time. For instance, you may have regulation that— Senator DAVID POCOCK: We don't have one. We're at the starting point. Prof. Flew : Sure. Senator DAVID POCOCK: Where do we start? Prof. Flew : Well, often these measures start with certain keywords as triggers, and that's a place to start, so a word like 'suicide' may constitute a trigger and that is embedded into the system, where there would be some sort of a blocking of interactions at that point. We do know that these things move over time. So you would need a regulatory framework that is sufficiently adaptive to that. Some years ago, I chaired the Australian Law Reform Commission's inquiry into media classification. What became apparent is the sorts of things that are a public focus change over time. The legislation was designed at a time when public swearing was a big issue. Now there has been a major decline in community concern about public swearing, but self-harm, information about body image, cyberbullying et cetera are now issues that are very central to our concerns whereas they weren't at the time at which media classification laws were originally developed. So there needs to be an adaptiveness to enable responsiveness to changes over time in what are the issues of concern. But there also needs to be some clear lines of accountability and some sort of mechanism. I think this is the challenge: who does it? If we're not leaving it to the companies themselves, are we talking about a co-regulatory model? Are we talking about a government regulatory model? The upside of a government regulatory model is legitimacy. The downside is that they're not particularly adaptive over time, and there would also potentially be concerns about overreach or censorship and so on. So co-regulatory models may be a suitable approach to be taken here. Of course, one of the challenges we have is that very few leading AI companies are headquartered in Australia. This is an issue with social media regulation and other forms of platform governance. But there can be conditions attached to operations. We do know that companies that are ostensibly global actually have the capacity to operate differently in national markets. I mean, it's a cost to them but it is possible. Senator DAVID POCOCK: How do you see that playing out? If one of these company's chatbots tells someone to commit suicide and they do, would that be a director's responsibility? Prof. Flew : In the first instance the trigger should be around the word 'suicide'. There should be something like a kill switch built into the framework around certain triggering words or phrases. The best first response is to try and stop that situation from emerging. Should it arise, this would be where terms of service conditions would start to become relevant. They would establish a degree of legal liability that would prompt the companies in question to be proactive around these issues before they arise. Were it to go ahead, is the company itself liable? Are the directors liable? How do you administer a fine? This is a question, for instance, with the social media age restrictions. The maximum fine has been increased from $49.5 million to $99 million, but we know that for these companies they're not big figures. Once you start trying to do share of revenue you get a range of jurisdictional disputes arising there. It's definitely challenging. There is some work—and I could provide this on notice to the committee— Senator DAVID POCOCK: That'd be great. Prof. Flew : that has been done recently on what has been called the adaptive state and the need to legislate in ways that allow for a degree of leeway for the agencies in how they enact this. I think, probably, in Australia the office of the eSafety Commissioner has been a leading example of that. Senator DAVID POCOCK: It seems like one of the things that OpenAI has offered the Australian government—I don't know if it's reparations or some sort of thing after this Medicare hack—is some credits through its billion-dollar Daybreak for Frontline Defenders program. Does that cut it and how should we— Prof. Flew : I wasn't aware of that. Senator DAVID POCOCK: Well, we don't know any of the details yet, but we covered it briefly yesterday. How should Australia be getting access to frontier models? Should that be through diplomatic channels—Five Eyes, AUKUS? It seems to me that I sit here, the US is supposedly our closest ally—we hear about that all the time—yet we're being told that we have to have training data centres here so we can get access to the frontier models and yet the president could turn around, as he has done, and say, 'These countries have no access to these models.' Surely, we should instead be saying, 'Okay, let's work diplomatic channels, have agreements with the US and have agreements with middle powers.' How do you see that playing out? Prof. Flew : I think share of market becomes important here. Australia is small but not insignificant in the global market. I think that often the likelihood of major companies moving out of Australia can be overstated. I think in the Asia-Pacific region, for the most part, there are largely two locations where global companies are headquartered; they are Australia and Singapore. There are very few other cases there. I think that there will be a process of embedding over time. This is where the opportunity arises to realise the AI dividend. Access to regular energy sources and stable compute power generates benefits for the companies involved that are enormous at this stage. The market may change, but at this point in time the benefits are enormous and I think there are opportunities to put some quid pro quos around that. Historically, I think about the car industry—the fact that US manufacturers developed the capacity behind the Australian tariff wall. In the broadcasting area, local content requirements became a condition for licenced operations. I think there is some capacity for some imagination here, and I also think that we should be looking at how some of that surplus can be invested in the cultural sphere. I think that the hurt that is felt among artists and writers—the many people involved in what we refer to as content creation, although I know that's not language artists necessarily like—is considerable, and I think that the capacity to create some sort of a fund to support creative work may be something to consider. To put something radical on the agenda, a basic income for artists may be something under consideration. I know that the challenge then starts to become how you define an artist, but, be that as it may, given that, under the framework that we have now, the average income of artists has barely moved over a 20-year period, I don't think just further defending the status quo is going to be sufficient. Senator DAVID POCOCK: Sure. We heard yesterday about the fund and the fact that maybe 90 per cent of that fund would have to be sent overseas to copyright holders, so I think there are a lot of questions to be answered about that model. But thank you. Prof. Flew : Sure. I know that it's what they call the 90-10 rule: 90 per cent of the revenue goes to 10 per cent of artists, because, if you're Elton John or Paul McCartney or something, you've got a lot of capacity to track what's being done with your work. Most don't. CHAIR: I'm conscious of time. We're right on time. I understand Senator Sheldon has a question, and then we'll move to our next session. Senator SHELDON: Thanks very much for presenting today to the committee. I will just go back to the question on trust. You were talking about the media. I will just be a devil's advocate in a sense. Why should we trust human trust rather than AI trust, in the case of the media? What's the answer to actually developing a conscious trust in human creativity and intelligence? As you've outlined, there's substantial loss of trust in institutions and organisations and increasingly in individuals. Prof. Flew : Yes. It's been tracked primarily through the Edelman Trust Barometer, though there are a number of other sources, such as the OECD and so on. Australia tends to feature in the middle of the list of countries around trust—to put it simply, we are below Scandinavia but above France, Russia and increasingly the UK and the US—with the exception of media. There is a very low level of trust in the media in Australia by international standards. Healthy scepticism has value in democracies—there's no question about that—but there's reference here to first-order and second-order trust. Typically, when people are asked about trust in the media, they often are referring back to their own media, so they might say, 'I trust the ABC, but I don't trust the Daily Telegraph,' or, 'I trust the Daily Telegraph, but I don't trust the ABC,' or whatever the framework might be. That's to be expected. In fact, it's to be valued in a pluralist democracy. It's the distrust of all sources of information that is the concern, and that's a question of the information ecosystem. That's what's in danger with a proliferation of AI generated misinformation and slop: people just don't believe any of it, which I think is different to scepticism. Senator SHELDON: Flowing from that, there's a broader question I thought you raised, and that is about people's trust in information they receive from AI on news queries. I will extend the conversation you're having with us: if there's more trust in using AI to diagnose health risks, to have a friend in need or to address psychiatric or social—psychosocial—injuries, or whatever people might think that they would get from an AI companion, how does that challenge get met? Are we meeting that challenge in a series of cultural ways at all yet, whether it be in the workplace or in the broader community? Prof. Flew : We're at very early stages of using AI as that form of source of information. I guess my point would be that AI ultimately doesn't generate information. This is all human sourced information. To the extent that, as the AI doomers argue, AI starts to become the generation of more AI, we will just go down a sinkhole of diminishing returns to information. I think AI system developers are aware of that danger. That's where different forms of a bargain can be reached around this, because AI companies ultimately have a stronger reliance upon expertise than may be apparent. CHAIR: I'm conscious of time. I understand Ms Chaney has one question. We might put the question and then take the response on notice. Ms Chaney? Ms CHANEY: I'm happy to put it on notice, Chair. CHAIR: Thank you, I appreciate that. Thank you, Professor Flew. We really appreciate you making yourself available for this committee and its inquiry. Please send any additional information you have taken on notice through to the secretariat in the next two weeks..