You expect to be safe when you go to work or when your kids play at the school playground. When you go to the doctor or get financial advice from the bank, you expect they’ll act in your interests.
In short, there are many places in our society where the people with more control or more power have a duty of care to the people using their spaces and services.
Digital platforms should be no different. They’re important places for accessing information and participating in our communities. They need to be safe and trustworthy.
This is why the Australian government is drafting legislation for a “digital duty of care”. It would require social media platforms and other online providers to establish risk management systems – to identify potential risks of harm from their services and take reasonable steps to prevent or mitigate serious harms.
However, we’ll only know if it’s working if digital platforms, governments and the public can meaningfully observe what’s going on.
This is known as platform observability. Our ongoing work and examples from overseas show that mechanisms for observability and transparency need to be, and can be, built into the regulatory framework.
Putting the responsibility on platforms
Australia already regulates online safety through three main mechanisms:
- the kids’ social media account ban
- complaints systems so people can report illegal or harmful content (like cyberbullying and harrassment) and request it be taken down
- codes of practice for limiting access to age-inappropriate content.
But these measures are inadequate when the algorithms that shape our online environment are promoting harmful content. Current product safety rules under the law don’t extend to online service providers, and it’s unclear at best whether or when general tort law would impose duties.
United Kingdom and the EU have already imposed duties of care on platforms. Australia plans to follow.
Australia’s digital duty of care should require platforms to take reasonable steps to mitigate harms from algorithms that hack our attention, promote scams, target us with ads for addictive products like gambling and alcohol, or show us harmful content about eating disorders or untested health products.
Platforms have more data and technical capacity than any single person or regulator to tackle these harms.
We also argue that a digital duty of care shouldn’t be just about reducing risks of harm. It should also entail reasonable steps to ensure information with public and community value – such as trusted public health information – isn’t buried or banished from our feeds.
How will we know it’s working?
Just as important as creating a duty of care is knowing if it’s working. Our feeds are highly personalised and content is often ephemeral, disappearing within hours. So we need to build in observability measures.
This can’t just be reporting, where platforms self-assess according to standards they largely define. Most transparency measures currently offered by platforms provide only partial information.
Our own research on digital advertising demonstrates this problem. Advertising libraries are incomplete, transparency reports aggregate away important detail, and user-facing explanations (“why am I seeing this ad?”) offer limited insight into the complex systems that shape targeting and recommendation.
Knowing why one person received a particular ad tells us very little about who didn’t receive it, or reveal patterns of exclusion, or unequal access to information.
To understand the broader scope of platform dynamics, we must observe systems at scale and over time.
A lack of independent access
Meanwhile, platforms have increasingly started to control who gets to research their content. And platform-approved researchers can often only see what platforms choose to make visible, in formats platforms define, using tools platforms control.
This challenge reflects a broader media transformation. In the 20th century, broadcasters distributed content, audiences consumed it, advertisers funded it, and independent ratings agencies measured what was happening.
This provided some independent verification. Advertisers could assess audiences, creators could understand reach, regulators could evaluate compliance, and the public could scrutinise claims because anyone could watch a broadcast.
But platforms now create the measurement systems, control access to the data, sell the advertising, curate the content, and report on their own performance.
Independent observability has become increasingly difficult.
A transparent ecosystem
A digital duty of care should be accompanied by an ecosystem of observability. Within it, researchers, journalists, regulators and civil society organisations could work together to understand how platforms operate, and hold them accountable to public values.
This would not replace regulation. It would make regulation possible, helping to identify emerging risks, evaluate whether mitigation measures are working, and provide evidence for policymakers. It would also give Australians greater visibility into the systems that shape their lives.
To achieve this, legislation should include three additional protections, based on our research experience to date.
First, researchers conducting legitimate public-interest work should be protected from platform retaliation.
Second, platforms should be required to provide meaningful access to data that enables independent research. This data should be available to be exported for research without limits on the publication of that research.
Third, Australians should have stronger rights to access, download and donate their own platform data for research purposes.
Together, these measures would create the foundations for ongoing accountability rather than one-off compliance exercises.
The digital duty of care is an important opportunity. But its success will depend both on scope of the duty, and on creating the ecosystem needed to observe, evaluate and challenge platform behaviour over time.
