By 2050, smart cities and homes will be the norm.

Your fridge will tell a drone to pick up

fresh milk when you run out, and lampposts

will change the intensity of street lights

when smartphones and therefore humans are nearby.

In this hyperconnected world, the threat cyber-criminals

pose will only grow.

In the home, smart devices, including voice assistants,

vacuum cleaners, and toilets, will

be easy pickings for hackers.

While these poorly secured gadgets do not

store sensitive data themselves, they

link to others that do and are vulnerable access

points for criminals.

Meanwhile, artificial intelligence

is a double-edged sword.

While AI systems can help to identify incoming threats,

attackers could also use them to unearth vulnerabilities.

In future, foreign powers or cyber-criminals

could cripple a country's electricity network

by taking over the AI that controls it,

or they could cut off the water supply

or even cause traffic chaos on the roads.

AI could also help identity fraudsters

by generating deepfakes.

These digital doppelgangers are currently a novelty,

but the techniques used to create them are rapidly

improving, with increasingly realistic depictions of people.

So far, they've mostly been used as part of revenge pornography.

But a realistic digital avatar could

be a useful cover for wheedling key details such as passwords

or bank details out of targets.

But the struggle to keep systems secure is not hopeless.

By 2050, much poorly written and outdated code

will be removed and replaced by more secure alternatives.

Even passwords could be phased out, obsolete and less secure

than facial recognition software.

Deepfakes could be beaten by constant verification systems

that track eye movement, faces, and keystrokes

to ensure the person behind the keyboard

is who they claim to be.

But that risks putting cybersecurity over privacy.

Policymakers and designers must work

hard to ensure that this does not become a binary choice.