I rise to speak in support of the Online Safety Amendment (Digital Duty of Care) Bill 2026. It is really urgent that we have this in place, and I want to thank at the outset so many who have campaigned for this and done a lot of work, particularly Chanel Contos with her Fix Our Feeds campaign, to give users back that control and autonomy of choosing what you see and what you switch out of in your feed. Many organisations have championed this because it's an area that is really the hallmark of this generation: how exposed they are to this online world. Yet it's an area that has been devoid of proper protections, so I do welcome the fact that we are finally here with a digital duty of care legislation.

It is time to rebalance responsibility for online safety in Australia. For too long, the burden of navigating online harms has fallen on individuals and families. We need to prevent the harm before it happens, rather than leaving individuals and families to deal with the consequences. The list of harms is long. When it comes to mental health, eating disorders, violence, sexual assault, intimidation and discrimination against women, we know that the online world has given rise to this environment where harm is allowed to go unchecked, and it's led to major issues when it comes to suicide and eating disorders. There is another area, though, that this bill does fail to address, and that's the growing threat of AI deepfakes, which use people's images, voices and identities without their consent. We know deepfakes can spread misinformation, fuel hate and damage reputations and public trust. Yet, despite the events—despite everything that we are discussing at the moment around rising threats and violent commentary online—this bill is failing to address what I would say is a major area that should be addressed.

We know the internet has transformed how we communicate, how we learn, how we do business and how we participate in public life. It has created extraordinary opportunities for connection, innovation and democratic engagement, but we also have to acknowledge the serious harms that can arise from these technologies. Social media platforms can amplify harmful material, recommend it to vulnerable users and spread abusive content at a scale unimaginable a generation ago. That's why it is so important that that control in relation to the feeds be given back to the user.

We can't expect individuals, especially young people, to bear the full responsibility for protecting themselves from highly sophisticated systems designed to capture and keep their attention—that's the really important bit. This is not innocent technology; this is highly sophisticated technology that is creating significant harm.

We can't leave families alone to deal with the consequences or navigate the risks alone. We've had so many stories of families absolutely devastated by tragedy as a result of the pernicious and damaging influence of social media—the algorithm and the pushing of harmful content—with no real action coming in return. Busy, exhausted parents should not have to rely on settings that are difficult to find and confusing to use or be sent through opaque and impersonal reporting and complaint processes that all too often result in very little change.

Technology companies have access to the data, expertise, resources and the technical capacity to identify and mitigate foreseeable risks at a scale that individuals simply do not. That is at the heart of the requirement of this legislation, and that is why it's important. They have the tools and the capacity. They must use those to mitigate foreseeable risks.

That's why this is important reform. It establishes a clear principle: online services should take reasonable steps to protect their users from serious harms arising from the use of their services. We accept that of every other business. For every other business, even if it's a supermarket that you walk into and buy a product from, we expect there to be compliance when it comes to duty of care—that a product can't cause harm. There are consequences when the risks are ignored. We know that, too often, big tech companies wait for harm to occur and only act if they're really forced to. Even then, it is a convoluted process. I welcome this move to a model where platforms must identify risks and prevent foreseeable harm in the first place.

The people of Warringah are highly digital aware. They appreciate the opportunities presented by digital technologies and AI, but they are also concerned about the harms. Since before this bill was introduced, my office has been contacted by many people who want to see greater protections in the online space. They want to see a digital duty of care passed. They believe that it will create a safer online environment, particularly for young people and vulnerable groups.

The age limits that the government proposes to pass are simply not sufficient—I've said that from the start—to keep young people safe online. In particular, young people in Warringah have told me loud and clear that they want online safety to be prioritised. A 16-year-old in the electorate wrote to me just a few days ago to urge me to support this bill in the hope that it will build a safer online world.

The push for a digital duty of care has been building for some time. Former member for Goldstein, Independent MP Zoe Daniel, in fact helped bring the idea into parliament, introducing her own private member's bill in November 2024. It would have required major online platforms to identify and manage the risks their services pose to Australians. The bill lapsed when parliament was dissolved in 2025, but the legacy continues. I welcome the Albanese government's developing this digital duty of care with the same principle at its heart: online platforms should do more to prevent foreseeable harm. I've long argued that online safety must involve meaningful accountability for the companies that design and profit from these systems. Remember, they are the ones that benefit from algorithms that keep pushing dangerous content.

But the details matter. We need legislation that delivers real safety and accountability and greater choice for users. Of course, everyone fearing this legislation has been hiding behind smokescreens of 'freedom of speech', 'freedom to choose' or 'government intervention'. In fact, I would argue that it's the complete contrary. This is giving choice back into the hands of users. They can control their feeds, and they can no longer be preyed on by these billion-dollar-making companies who benefit and profit out of pushing content through these algorithms.

We know that as the technology evolves so must our approach to the harms it can create. We have the opportunity in this legislation to ensure that it addresses not just the risks that we already understand but also address emerging threats. One such risk is the spread of highly realistic AI generated audio, images and video depicting identifiable people without their consent. I've been trying to get the government to engage on this when it comes to truth-in-political-advertising legislation and to cover deepfakes in that, but it goes broader than just political advertising. It's actually everyone. Why should anyone in Australia have their image or likeness used in a way that is contrary to their wishes and without their consent to do something that can be very harmful? The amendment that I intend to move will address this issue.

Artificial intelligence can make someone appear to say something they never said. It can create a scene that never happened or make a fabricated event look like a genuine news report. We do not have to look far to see examples of these risks in Australia. Consider the AI generated videos released by Sportsbet featuring high profile AFL players Lachie Neale and Will Ashcroft. The videos depicted the athletes saying things they had not said and were made without their consent. The videos were reportedly intended to help Sportsbet attract new gambling customers, and the AFL instructed Sportsbet to take them down. They were able to do that because they had that power, but an individual in the community does not have that same power to ask for a deepfake to be taken down.

A deepfake video depicting Tasmanian Senator Jacqui Lambie has also recently circulated on social media. It portrays her making statements that she never made, including comments criticising another party's voters, which then led to increased threats against her because they then were outraged by the perception that these comments have been made. These are the risks we face and the dangerous environment in which we are in. These examples raise a fundamental question should someone or an organisation or an account overseas be able to use artificial intelligence to put words in another person's mouth and use their identity, image and likeness to deliver a message they never agreed to deliver? I think any right-minded Australian should say a firm no to that.

AI can also be used to create fabricated scenes that appear realistic. Earlier this month, the One Nation Victorian branch aired an AI generated video featuring a man of African appearance wielding a machete and a suicide bomber who appeared to be of Middle Eastern descent. This is incredibly dangerous content. The video used fabricated imagery to portray threatening scenarios, stoke hate, provoke fear and create a highly racist and dangerous environment. It demonstrates how synthetic content can influence public perceptions, particularly when viewers may not realise that the scenes are artificial. As AI tools improve, the distinction between authentic and synthetic content becomes increasingly difficult for ordinary viewers to make. This material can also be produced, copied and distributed at an extraordinary speed. The consequences can extend well beyond the original post. They can affect personal safety, professional reputations, public trust and the quality of our democratic debate. Australian law must respond to these risks in a way that is practical, proportionate and capable of adapting as the technology develops.

The European Union has introduced transparency requirements intended to make AI generated and manipulated content easier to identify. Since August this year, the EU's AI framework has required certain deepfakes to be disclosed as artificially generated or manipulated. The framework also includes requirements concerning machine readable markings of some AI generated content. These European laws recognise that people should not have to guess whether what they are seeing or hearing is genuine. The European approach also recognises that art, fiction and satire should not automatically be treated in the same way as content pretending to be authentic. Transparency requirements are not the same as a blanket ban on deepfakes. What we should do is seek to avoid people being deceived while allowing legitimate uses of AI to continue. We need to look at that, and the framework, to recognise the benefits of innovation while ensuring that people affected by fake content are not left without meaningful protection.

We need to ensure that deepfakes which depict an identifiable person without their content and are likely to cause serious harm are captured within this bill. That means content that depicts you, Madam Deputy Speaker Garland, and any other member in this House but also members of the public. We should not have the right to use someone's image, voice or likeness in AI generated content without their consent. It's not about prohibiting artificial intelligence in creative works or preventing Australians from commenting on public figures, but we cannot pass that off as being authentic content. We need to address the specific risks where AI generated content that uses a real person's likeness without consent is capable of misleading viewers into believing it is genuine and is likely to cause serious harm, not just misinformation. I commend my amendment to the House as incredibly important if we genuinely want to make this environment safe.

Social media platforms are responsible for assessing and managing risks associated with this material. This would force them to consider and take reasonable steps to reduce that risk. It would also give the eSafety Commissioner the power to order the removal of harmful online content. Platforms shouldn't wait for a victim to suffer consequences before taking reasonable steps to address a foreseeable risk. It's particularly important because deepfake content can spread so rapidly. In fact, we saw it in the aftermath of the Bondi terror attack, where we saw fake content generated online; that should be banned. Once a video or image has been widely shared, removing the original material may not be enough to undo the damage.

The bill recognises that AI generated abuse is a serious and growing problem. The proposed powers concerning fake nude material are an important example of the need to act before harm spreads, before the person is depicted, because it's so devastating. But we need to go beyond just that because the harms are much greater. I support the principle, but I ask the government to look for all AI generated fakes. Many are worried about freedom of expression, which is front of mind. Satire and political commentary are a legitimate and valuable part of democratic life, but they are not a licence and we shouldn't be creating a framework that enables us to abuse and take on the image, likeness and content of another person. I commend the bill to the House in relation to the digital duty of care, but I urge the government to go further.