An 'off switch' for algorithms is not enough. Tech companies need a legal duty of care
Governments keep reaching for easy fixes: ban kids from social media, add an opt-out button. Campaigners say that misses the point entirely, and the real target should be the addictive design baked into the platforms themselves.
Key points
- No government has yet passed a binding digital duty of care law that holds tech platforms legally responsible for harmful design choices.
- The Australian under-16s social media ban, cited as a model by other governments, is widely described by researchers as unenforceable and easy to circumvent.
- Letting users opt out of algorithmic recommendation, the system that decides which posts and videos you see next, treats a symptom rather than the cause.
- The core demand from campaigners: make platforms legally liable for the safety of the environments they build, not just the content that appears inside them.
When a government wants to look tough on social media, it reaches for the same two tools. Ban children. Or give users a switch to turn off "the algorithm." Neither one actually changes how the platform is built.
That's the argument gaining ground among digital-safety advocates, laid out sharply by Zoe Daniel writing in The Guardian AI. The target, they say, should be the design itself: the infinite scroll, the notification drip, the recommendation engine tuned to keep you watching rather than to keep you well.
What is a digital duty of care, exactly?
A duty of care is a legal obligation, already familiar from medicine and construction, that requires someone to take reasonable steps to avoid harming the people they serve. A digital duty of care would apply the same principle to tech companies.
Under such a law, a platform could be sued or fined not because a specific piece of harmful content appeared, but because its product was designed in a way that foreseeably caused harm. That's a much higher bar than anything currently on the books in most countries. We first covered this accountability question on 19 July 2026, when Australia was drafting rules on how government AI can affect ordinary citizens.
Why do critics say the social media age-ban failed?
Australia's law banning under-16s from social media is the most discussed recent attempt. It looks decisive. In practice, a teenager with a parent's email address and a VPN, software that hides your real location online, can be back on a platform in about three minutes.
The ban doesn't touch the product design that makes these platforms hard to put down. Advocates compare it to banning children from a poorly wired building without requiring anyone to fix the wiring.
| Policy approach | What it targets | What it leaves untouched |
|---|---|---|
| Under-16s social media ban | Access by age group | Addictive design, algorithmic ranking |
| Algorithm opt-out button | Recommendation for one user | Default settings for everyone else |
| Content moderation rules | Specific harmful posts | The design that spreads them |
| Digital duty of care law | Platform design and business model | (Intended to cover both) |
What happens next?
No major government has passed a full duty-of-care law yet. The UK's Online Safety Act moved in that direction but stopped short of making platforms liable for design decisions. Campaigners are pushing for the next legislative cycle to go further.
Here's what I'd tell any reader watching this space: the next time a politician announces a new social media policy, check whether it changes what companies are legally required to build, or whether it only changes what you're allowed to do inside what they built. Those are very different things, and most recent announcements are firmly in the second category. The platforms know it too.
