AI glasses solved four university-entrance-style maths questions in ten seconds in a Korea Times demonstration, sharpening the challenge facing South Korea’s College Scholastic Ability Test (CSAT). The bigger question, as the exam approaches, is whether better security can keep pace with AI that may not need an internet connection.
Watch Desk analysis
What happened
A Korea Times opinion article describes a recent tax-accountant exam cheating case involving a miniature camera, AI and a smartphone, and says the suspect went undetected until an unusually high score drew attention. The article argues that emerging devices could make cheating harder to spot. It also notes that Google’s Gemma 3n can process images and text on-device, so cutting internet access would not necessarily prevent AI use.
The article points to Cornell’s EchoSpeech and InvisiSpeech, a research system using a sensor near a back tooth, as examples of technology that can detect silently mouthed commands or subtle vibrations. These are research examples, not ready-made exam-cheating systems. The piece also reports proposals to move towards absolute grading and written responses, while stressing that any replacement must earn students’ and parents’ trust. Read the Korea Times article.
Why it matters
The immediate issue is exam fairness: AI-assisted cheating could be difficult to prevent with familiar measures such as restricting network access. But the article’s more enduring point is that a test built around selecting answers under time pressure may be a poor match for a world where tools can produce answers quickly. Asking students to explain their reasoning or identify errors could test knowledge in a different way.
Security and assessment design are separate problems, though. A harder-to-cheat exam still needs to be fair, credible and trusted by the people whose futures it helps decide.
Our read
The ten-second demonstration makes a crisp headline, but it does not settle what an AI-proof exam should look like. The useful argument is about what schools want to measure: getting the answer, or understanding why it is right. Better security may buy time; it cannot decide that question for educators.
What to watch
- Whether South Korea’s exam authorities change rules or screening ahead of the CSAT.
- Whether proposed written-response or absolute-grading reforms gain support from students and parents.
- How exam designers distinguish real-world research prototypes from devices that are actually available to candidates.
Discussion spark: Should high-stakes exams prioritise stronger anti-cheating controls, or shift towards written answers that show how students reached a result?
Sources and evidence
- Source update (16 November 2026, 07:15 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.