GP / Blog / Science & Technology

Issue forecast

Surveillance, privacy and the bargain no one signed

Surveillance is the strongest sleeper in the bank. It last appeared in 2009, when the camera was a fixed thing on a pole and the phone in your pocket did not yet track you. The technology has been rebuilt twice since. The question has not been asked once.

Theme · Science & TechnologyLast set at A-Level · 2009 (on technology, privacy and security, and largely untouched at A-Level since)Foresight, not a prediction

In short

A surveillance question rewards a student who refuses the lazy frame of 'privacy is dead, get over it'. The real argument is about consent: we have a vast apparatus of cameras, face recognition and data collection that no one was ever asked to approve, justified each time by safety or convenience. Argue whether that bargain is worth it, and whether we ever actually made it.

Why this could come up now

This is the longest-dormant strong issue on the slate. The theme map shows privacy and surveillance last set in 2009, before the smartphone era, before social media saturation, before facial recognition was deployable at scale. Over a decade cold at A-Level while it dominates the news is exactly the pattern that makes a sleeper worth rehearsing.

The current affairs make it loud. Through 2025 and 2026 a US border agency contracted a face-recognition firm with access to over 60 billion scraped images; American states scrambled to write facial-recognition law in the absence of any federal one; the EU moved to tighten biometric rules. The apparatus is expanding faster than the rules around it.

Framed honestly: a question cold since 2009 is not a prediction that it returns this year. It is the clearest case on the slate for preparing an issue most candidates have never thought to argue, so that if it comes, you are not improvising a topic the bank has not touched in your lifetime.

200920122015201820212024

set at A-Level most recent appearance. Cold since 2009 at A-Level while exploding in the world. The definition of an overdue sleeper.

What an essay on this would test

A surveillance question tests whether you can weigh a genuine trade rather than strike a pose. Cameras and data really do catch criminals and speed your life along; they also build a record of you that you never agreed to and cannot inspect. The strong answer holds both and argues where the line should sit, not which side is the villain.

It also tests whether you can localise. Singapore is one of the most camera-dense, data-integrated societies on earth, and its citizens have largely accepted that bargain in exchange for safety and order. A Singaporean student who argues surveillance only in the abstract of Western dystopias misses the more interesting case sitting at home.

Operative angles

privacy versus security: the classic trade, but the question often turns on whether it is a real trade or a false choice
consent: the heart of the contrarian case, since most data collection happens without a meaningful decision by the person collected from
convenience: the quiet driver, since much surveillance is accepted not for safety but because it is easier

How to answer it: two ways in

Two distinct, defensible routes through the question. A strong script commits to one and uses the other as the concession it answers, rather than sitting on the fence.

The bargain is worth it

Safety and convenience we would not give back

Surveillance, in moderation and under law, buys real goods, lower crime, faster services, found missing persons, and most people accept it because the trade is genuinely favourable, not because they are dupes.

Cameras and data deter and solve crime, and few who benefit from a safe street would trade it back for total anonymity.
Much of what we call surveillance is convenience we chose: contactless everything, instant identity checks, services that remember us.
Singapore's broad public acceptance of dense monitoring, paired with low crime and high order, suggests the trade can be reasonable, not only sinister.

Worked exampleDuring COVID, Singapore's TraceTogether contact-tracing system reached most of the population and aided a response that produced among the world's lowest case-fatality rates, a concrete case where giving up some location privacy bought a public-health good (Singapore government, as of 2026-06).

We never consented to it

A surveillance apparatus no one was asked to approve

The deeper problem is not whether surveillance is useful but that the apparatus was built around us without consent, expanded by each agency one safety justification at a time, until opting out is no longer possible.

Facial-recognition databases are often built from images scraped without permission, so the data subject never agreed and cannot withdraw.
Each step is justified by safety, but the steps accumulate into a capacity for tracking that no one voted for and no court fully oversees.
Privacy is not just personal comfort; it is the space in which dissent, mistakes and ordinary freedom survive, which is why losing it by default, not by choice, is the real danger.

Worked exampleIn February 2026 the US Customs and Border Protection agency contracted the facial-recognition firm Clearview AI, whose system draws on more than 60 billion publicly scraped images, for tactical targeting, an expansion of biometric surveillance built on photographs the people in them never consented to being used this way (FedScoop and Biometric Update, Feb 2026, as of 2026-06).

The fuel: stats, facts and examples

60bn+

publicly scraped images in the Clearview AI system contracted by a US border agency in Feb 2026

Source: FedScoop / Biometric Update · as of 2026-06

13 + 23

US states and local jurisdictions with laws specifically addressing facial recognition by late 2025, with no federal law at all

Source: Security Industry Association guide, Dec 2025 · as of 2026-06

Jan 2026

the Texas Responsible AI Governance Act took effect, restricting use of biometric data scraped from public sources without consent

Source: Texas Responsible AI Governance Act (signed Jun 2025) · as of 2026-06

Feb 2025

when the EU banned most real-time remote biometric identification in public spaces for law enforcement, with narrow exceptions

Source: EU AI Act, Article 5 · as of 2026-06

Facts worth deploying

There is still no federal facial-recognition law in the United States, which is why a patchwork of state and city rules has grown up around a technology federal agencies are themselves adopting.Source: NPR / Security Industry Association, 2025, as of 2026-06

The EU's AI Act prohibits most real-time remote biometric identification in public spaces for policing but treats after-the-fact biometric analysis as high-risk rather than banned, a line that itself shows how contested the rules remain.Source: EU AI Act, Article 5, as of 2026-06

Singapore's smart-nation infrastructure, dense public cameras and the COVID-era TraceTogether system, makes it a live local case for arguing whether a high-trust society can accept surveillance that others would refuse.Source: Singapore government, as of 2026-06

FAQ

Is it safe to criticise surveillance in a Singapore exam?

Yes. GP rewards balanced, evidenced argument, not a political line. You can credit Singapore's safety bargain genuinely and still scrutinise the consent problem, which is the same move a strong essay makes on any state policy. Argue, source, and stay fair to both sides.

How do I keep this from becoming a list of scary technologies?

Anchor every example to the argument, usually consent or the safety-versus-privacy trade. Do not catalogue cameras and face scanners; ask, for each, who agreed, who benefits and who is exposed. The technology is the evidence, not the subject.

Why is the consent angle stronger than 'privacy versus security'?

Because security-versus-privacy invites a tired both-sides shrug, while consent sharpens it: even a surveillance system that works can be illegitimate if it was imposed without choice. Arguing consent lets you concede that surveillance is useful and still hold that the way it was built is the real problem.