"The draft duty of care legislation includes two key elements that warrant further attention, and one that is missing entirely. Firstly, the draft legislation would require platforms to take reasonable steps to ensure children are protected from harmful content. While this is much needed in digital spaces, there can be a danger that platforms may opt to use over-compliant content moderation systems that result in harm-reducing content being caught up as false positives.

For example, automated systems could block or remove media literacy content created to help people identify problematic content that glorifies hostile attitudes towards women, or content created to support people seeking help with disordered eating or bullying. Secondly, the draft includes a requirement for platforms and services to develop “user empowerment tools”. This has been framed as an ability to either opt in or opt out of an algorithmically curated social media feed, yet all-or-nothing approaches don’t leave much room for nuanced user preferences that might include a desire for limited recommended content alongside content from accounts that are followed.

It is promising though that as drafted, nuance could be a required principle of these “user empowerment tools”. Finally, the draft does not explicitly or implicitly account for how recommender features promote content that users themselves upload, including children. A true digital duty of care would explicitly ensure that users are also empowered to exert some control over who their content is being recommended to."

"The policy seems to address platform features but may underestimate the social nature of people’s online behaviour, which is largely socially constructed and influenced by other users.

Harmful online experiences are not produced solely by algorithms or platform design. Harmful online experiences can also emerge through peer influence, social norms, herd behaviour, online communities and users' own choices.

Changing the feed algorithm may reduce exposure to some harmful content without addressing the underlying social processes that cause users to engage with, share and amplify that content. The law may need to combine platform accountability with digital literacy and user education.

My Feed, My Way may not necessarily give users greater agency but simply gives them greater control over one component (algorithm) of a much larger digital ecosystem."

"The Digital Duty of Care is a step in the right direction, putting the onus back onto platforms to ensure they systematically address known issues, particularly in relation to algorithms and harmful content.

While users will have the option to opt-out of algorithmic feeds, it will be interesting to see what that looks like in practice – for instance will this be a one off event and what kind of communication will there be to help users understand what the options mean for them?

We know children and young people access algorithmically driven platforms, whether they have access to social media or not. In addition to the Digital Duty of Care, further media literacy education for children and young people is required to ensure they are equipped to make choices that suit their own interests and needs when engaging online."

"It is encouraging to see moves to give users more explicit control over their social media feeds. The recommender systems that curate our social media feeds are responsive to a range of user signals, as well as the business interests of platforms and advertisers, so user practices and preferences already shape much of what they see and hear online, but currently we can’t see or control the effects of these signals.

While many popular platforms have at times offered user controls, including an opt-out mechanism for algorithmic feeds, these have generally been difficult to find and use, often buried several layers deep in settings menus. The proposed ‘My Feed, My Way’ opt-out and/or opt-in mechanism has benefits in that it would require platforms to offer users a clear and explicit choice. However, as currently described, it sounds as though users might be offered only a very simple on-off switch, essentially enabling them to turn off the recommender systems and revert to a chronological feed.

It is uncertain how this would mitigate issues around so-called ‘filter bubbles’ and ‘rabbit holes’, for two reasons: first, because the choice to follow only accounts you already follow will tend to further reduce the diversity of perspectives we see; and second, because it is unlikely that most users will stick with a non-algorithmic feed that fails to engage them, and may choose to revert to the algorithmic version. It will be important to continue work on pressuring platforms to offer more safeguards, transparency and control over the feeds we already have."

"The “My Feed, My Way” initiative to give users some control of their feeds is a reasonable starting point to improving online safety. It places more responsibility on big tech and platforms; provided that the end users are actively aware and informed of the initiative.

Some users might not like personalised feeds and would prefer their algorithms to be tailored to their own needs and circumstances. Someone who is facing an overwhelm of content online might not want to be exposed to content that they did not seek out themselves, to protect their mental health. For others, for example those who have relocated interstate or overseas, might prefer recommendations which are relevant to their new information environments.

This move might help users understand the way people relate to their algorithmic feeds, beyond just on social media platforms.

The ability to take control of their feeds helps them move from being 'passive consumers' of online algorithms to being more 'active and empowered' users."

This comment was put together with collaboration from Dr Marc Cheong, Deputy Director of the Centre for AI and Digital Ethics (CAIDE) at The University of Melbourne

"The proposed “My Feed, My Way” initiative is a step towards giving Australians more control over what they see online. Through my research into consumer vulnerability, I have repeatedly found that emerging technologies and digital environments can shape and at times intensify harm or influence over social media users, implicating them psychologically, socially, financially and even physically. While the initiative recognises certain people may be more at risk of these harms, particularly children, it is important to note vulnerability can be far more expansive. Certain circumstances can make any of us more susceptible to harm, which is something Australia, and the global economy more broadly, increasingly needs to recognise.

Research, including my own, has identified a particular area of scrutiny in how algorithms can shape and sometimes intensify vulnerability through the content they repeatedly serve us. Social media algorithms learn from our behaviour online: what we click on, watch and engage with, and ultimately feed us more of that content. This can create echo chambers that reinforce particular views or behaviours. For someone already vulnerable, repeated exposure to precarious content can potentially deepen that vulnerability.

We are increasingly recognising that vulnerability isn’t always something a person brings to social media, but can also be created or amplified by the platform itself. While allowing people to 'switch off' certain recommendations could provide an important circuit breaker, there are caveats. We need to be careful that 'giving consumers control' doesn’t become another way of making consumers responsible for protecting themselves, or inadvertently allow consumers to create their own echo chambers.

The people most vulnerable to harmful algorithms may also be the least likely to recognise what is happening or change their settings. Real consumer protection requires individual choice and platforms taking responsibility for the environments they create, but also ensuring we deepen our understanding and provide education into digital vulnerabilities across social media users, businesses and broader institutional forces."

"Somewhere in the next year, every Australian on social media will get a notification asking whether they want the algorithm or their friends. The Prime Minister says this is about giving people control, not government. He is half right. The government has handed Australians a switch. Whether anyone flips it is another matter.

A 2025 survey of 6,217 people across six countries found only one in five would switch off personalised recommendations if offered, and a further quarter did not know. Europe has required this choice since 2023. Two years on, the feeds are hard to find, poorly explained, and users who try them drift back. Meta's own 2020 experiment showed a chronological feed cut time on Facebook and Instagram significantly. The algorithm is good at its job, and the minority who do opt out will find a feed of friends and creators feels quieter than the one they left.

Meta and TikTok know this better than anyone, which is why the wording, timing and design of that notification will decide whether My Feed, My Way delivers real choice or a box-ticking exercise. The bill needs to be specific about that, and the eSafety Commissioner needs the power to test it. The detail belongs in the legislation, not in Meta's hands."

"A digital duty of care is a legal duty, familiar from other products and services, from food to toys to playgrounds. Underlying this is an ethical principle – those who wish to sell and profit from products and services must protect those who use them from harm by making sure they are safe.

The duty of care approach is familiar to us from other fields – it has long been the case that products from cars to food to toys have been governed by regulations requiring vendors and providers to ensure they are safe before providing them to the market. The proposed new laws apply this to digital spaces: those who provide and profit from virtual spaces need to make them safe.

All of this is further underpinned by a very old ethical principle – the idea that ‘ought implies can’. This means I can’t be required to do something unless I am able to do it. This matters because it is the owners and providers of online platforms who are able to build them in a way that protects from harms. Individual users, who are less technologically literate and less capable of making changes to technology, are not able to do so.”

“Young people’s experiences of social media vary enormously. We need to understand what they are seeing, who they are connecting with and how those experiences affect them.

Research I’ve been involved in has highlighted how social media can help LGBTQ young people find support and explore their identity. Other work has examined young people’s experiences of fitness content and the pressure they feel to look a certain way. Any approach to making social media safer needs to account for both the benefits young people get from it and the harms they can encounter.

Giving users the option to remove personalised recommendations from their default feed could offer more control over what people see. But we need to test whether people understand and use that choice, and whether it reduces harmful exposure. Having a setting available is a great first step but does not tell us whether it works.

Platforms also need to take responsibility for harmful content and design features. We cannot expect young people and parents to manage those risks on their own.

Young people should be involved in developing and evaluating these changes. We need to hear what they want to avoid online, what they value and whether the changes make a difference to their everyday experiences.”

"The Digital Duty of Care may finally push responsibility upstream, from individual users towards the companies that design and profit from these systems. But Australia should look closely at the EU, where the Digital Services Act sets a stronger benchmark on independent oversight: major platforms face annual independent audits, systemic-risk obligations, data access for vetted researchers, and requirements around non-profiled recommendations. Australia has an opportunity to build on that model, but on the details currently public, parts of the proposal risk becoming a watered-down version.

The root cause of the problem remains further upstream. Algorithms do what the system around them rewards. Under the dominant platform business model, that means attention, engagement, and retention because those drive revenue. Meaningful reform therefore needs independent scrutiny, transparency, and enforceable public-interest obligations, not just additional controls for users.

That is also where “My Feed, My Way” needs more teeth. Personalised recommendation should be something users actively opt into, with a non-profiled feed as the default. The current announcement promises a choice, but does not clearly establish that protective default. Without an opt-in default, we are still shifting responsibility (and blame) for systemic harms onto individuals.

The direction is right, but Australia can be more ambitious. Choice is useful, but not a substitute for accountability."

"This move by the Albanese government sends a clear message to digital service providers that individual users should retain agency over the content they are exposed to online, and that they should be made aware of potentially harmful content. As a body image, eating disorder, and social media researcher, I will be interested to see to what extent digital service providers must demonstrate transparency over how they determine what is harmful content, and also the extent to which experts (for example, researchers and health practitioners) may be involved in making those decisions.

For example, when it comes to content that promotes or encourages eating disorders and body image dissatisfaction, this may cover a range of content from the fairly obviously risky extreme dieting and appearance modification behaviours such as looksmaxxing through to wellness and fitness inspiration which may appear somewhat innocuous but which research has shown can contribute to body and eating concerns, particularly for at-risk individuals.

I will also be interested to see how the changes are communicated to users by the Albanese government and the digital service providers, because the framing of these changes is likely to significantly impact whether individual users will opt in or opt out."

"The Albanese Government’s Digital Duty of Care legislation is a welcome and long overdue correction to a system that has allowed tech companies to make billions while manipulating our attention, shaping our behaviour and harming our mental and physical health.

For years, these platforms have engineered addictive features that keep children and adults online longer than they intend, often at the expense of sleep, wellbeing and real-world connection. This legislation finally acknowledges that the design of these products, not the users, is the problem.

The “My Feed, My Way” initiative is a meaningful step toward restoring autonomy. Allowing Australians to opt out of algorithmic feeds is not about restricting choice; it’s about reclaiming it. Algorithms have quietly dictated what we see and how we feel for too long.

Extending the Duty of Care to games, apps and AI systems is equally important. Neuroscience has shown repeatedly that developing brains are highly susceptible to high-stimulation digital environments, yet tech companies have faced almost no consequences for exploiting this vulnerability.

The fines attached to the legislation, while symbolically significant, are far too small for companies whose annual revenues dwarf national budgets. If we truly want accountability, penalties must be large enough to change behaviour, not simply absorbed as a cost of doing business."

"The draft legislation has some interesting features: I specifically like that it identifies certain design features that can have a negative impact on children (like an 'endless feed' feature, or a recommender feature, or 'time-limited' feature). I like that it directly addresses the fact that we don’t currently get to choose if we see information that has been algorithmically curated or not, but that we should all be able to choose.

It might have the side effect of forcing Meta (and ByteDance et al) to implement the original social media ban properly rather than stalling. Eg: they will face double penalties if they continue to let under-16s log in and see the design features listed above. Penalties here and penalties under the ban. It wouldn’t be good for their business to continue dragging their feet."

"Australia's proposed Digital Duty of Care represents an important shift in how we think about online safety. It compels those who design and operate digital services to identify foreseeable risks and take reasonable steps to pre-emptively address them. This is a much more mature approach to platform regulation.

Rather than applying only to the largest technology companies, the duty extends across all online service providers operating in Australia, with what is expected of them varying according to their circumstances and the risks their services create.

The comparison with workplace health and safety is a useful one. We do not expect workers to individually manage every risk created by their workplace; rather we expect those responsible for designing and operating that environment to understand and mitigate foreseeable harms. The same principle should apply online.

Importantly, the exposure draft also recognises that a duty of care cannot rely on platforms marking their own homework. The provisions for independent researcher data access, transparency and protections for public-interest research begin to give Australia's research community much-needed and long overdue capacity to independently observe how digital services actually operate and work in practice.

The detail will matter, particularly the rules that follow. But the underlying principle is sound. Responsibility for online safety should extend upstream to those who design, operate, and profit from our digital environments."