Agents of shield: What it takes to protect your smartphone (2025)

ByVishal Mathur

By now, our smartphones know us better than most of the people in our lives ever will.

PREMIUM

At the forefront of this fight are smartphone operating systems themselves.

Apple’s iOS and Google’s Android have evolved into watchful gatekeepers that track app behaviour and usage trends as part of an effort to prevent unauthorised access.

Behavioural analysis programs, for instance, use information they have gathered on location and usage patterns to flag possible “dark patterns” that might indicate an attempt to infiltrate a device. If such an attempt is detected, the operating system can lock out a suspected app and warn the user about a possible malware attack.

Apple’s tightly controlled iOS uses machine learning to track even silent attempts to ping location data to an external source. When public personalities are told their iPhones had been targeted for attack, it is often this feature at work

Android’s methods are, necessarily, more collaborative. Since Google has no control over the hardware choices that its clients, the phone manufacturers, will make, its Android systems cannot use Apple’s hammer-like approach.

How does protection work here? For the past decade, Android has collaborated with third-party security and VPN (or virtual private network) apps, which are themselves evolving rapidly. Once used merely to mask IP addresses or block trackers, they now use quantum encryption to detect the faintest hints of AI-led interference and other anomalies.

Quantum encryption uses particles such as photons to transmit information (rather than digital bits of 1s and 0s). Since these particles exist in multiple states, any attempt to intercept them disturbs their patterns, making incursions easier to detect.

Until recently, however, quantum encryption was so expensive and complicated that it was reserved for diplomatic cables and satellite transmissions. With quantum chips now more easily available, and far more efficient, these systems are being deployed within apps such as Proton VPN, NordVPN and ExpressVPN.

Such apps, as a result, can detect anomalies in voice calls too. If a scammer’s voice has been synthesised using AI, quantum encryption software can detect micro-discrepancies in cadence, acoustic patterns and spectral signatures, to raise an alert or even block the call midstream.

An ongoing duel

As with all things tech, the weapon for the defence also soon becomes the tool of the offence. Quantum-resistant algorithms will emerge, and companies are already testing new weapons they can use in the next phase of this fight.

Meanwhile, the other side is preparing too. “Threat actors are already swiping lots of encrypted information, often the kind with a long shelf life, and stockpiling it until quantum computers can crack it open for them in the future,” says Pete Membrey, chief engineering officer at ExpressVPN.

Privacy concerns, therefore, loom large. To keep encrypted data safe, it must stay protected; yet the only way a protection system can work is for the user to feed it with sensitive data of one kind or another. This could include data on app usage, or actual communications.

Can there be a better way?

There is already talk of AI being used to build smartphone security algorithms that use predictive policing. Such a system could potentially analyse the user’s behaviour, habits and usage to indicate when a threat might be imminent, based on local and global threat patterns.

The challenge here, of course, is accuracy. Predictive policing software has shown us the limits of algorithmic fortune-telling: these systems often amplify existing biases and generate more false positives than actionable insight. The future rarely unfolds as neatly as data patterns suggest they will.

Still, the arc could bend from reactive defence to proactive protection. For now, for better or worse, we’re simply not there yet.

See Less