Skip to content
Crossbencher.

Essay · Crossbencher ·

What would make children safer online?

A popular aim doesn't establish that a technical measure works. Online safety needs honest answers about effectiveness, privacy and enforcement.

Aswin Mahesh · Unsplash

Protecting children online is an aim almost everyone can support. That makes it particularly important to ask how a promise will work, rather than treating agreement with the aim as evidence that the mechanism is sound.

Commercial platforms have responsibilities. Families can't reasonably be expected to solve every problem created by services designed by specialist teams to hold users' attention. Government has a legitimate role in requiring companies to reduce harm and enforcing those duties.

The difficult questions concern what a system can detect, how reliably it can do so and what access it needs to private information.

Be specific about the technology

Public posts, private messages, known abuse images and newly created material present different detection problems. We shouldn't discuss them as though a single tool can identify everything with the same accuracy.

End-to-end encryption restricts what a platform can inspect as messages pass through its service. Scanning material on a device before encryption raises a different set of privacy and security questions. Describing that as happening before encryption doesn't, by itself, establish that private communication remains protected.

Equally, it would be too sweeping to say that every form of automated detection requires reading every message, or that every researcher agrees on every proposal. The actual design needs examining. What is inspected? Who controls the criteria? What happens when the system is wrong? Could the capability be repurposed?

What would count as delivery?

Passing a law and publishing a consultation are steps in a process. Neither is sufficient evidence that children are safer. But a consultation isn't proof of bad faith either. A complex duty needs workable standards before it can be enforced fairly.

The useful test is whether the government can explain the obligation, the technology and the evidence of effectiveness, including the consequences for people who aren't doing anything wrong.

Criminal charges against executives aren't the only measure of success. Prevention, effective investigations and reliable enforcement matter more than producing a dramatic prosecution on an arbitrary deadline.

I want a government willing to acknowledge the trade-offs and defend the choices it makes. A promise to protect children shouldn't be used to avoid answering questions about whether a measure will work or what powers it gives the state.

What would make children safer online? · Crossbencher