Policy & Regulation

Lee orders review of CSAT extension after emergency alerts

President Lee Jae Myung asked officials to consider compensating Thursday’s CSAT for emergency-alert interruptions while his government weighs random alerts to detect phones used in AI-enabled cheating.

Lee orders review of CSAT time extension over disaster alerts
Lee orders review of CSAT time extension over disaster alertsAlexas_Fotos / pixabay
By James Calloway6 min read

Why it matters

  • Late Tuesday, Lee asked the government to consider extending Thursday’s nationwide CSAT end time by time lost to qualifying emergency-alert interruptions.
  • The qualifying response requires a proctor to intervene and remove the test-taker from the room.
  • Lee said the necessary measures should take two to three minutes unless multiple cheating cases occur.
  • The government announced a ban on electronic devices in test rooms and a pre-exam emergency alert to detect unreturned phones.
  • Lee also asked officials to consider random alerts during the exam while chairing a video meeting from Manila.

By James Calloway

President Lee Jae Myung instructed his government late Tuesday to consider extending Thursday’s nationwide College Scholastic Ability Test (CSAT) by the time lost to emergency alerts.

The proposal would compensate test-takers when an alert makes a communication device sound, a proctor intervenes, and the candidate must leave the testing room. Lee described the adjustment as an extension to the exam’s end time.

The policy trade-off is concrete: the government selected disaster alerts to help prevent artificial intelligence (AI)-enabled cheating, while the president considered how candidates could recover time lost when those alerts cause a device to sound. That is the concern raised in the post Lee shared.

Lee disclosed the instruction in a late-Tuesday post on X. He shared another user’s post expressing concern that loud disaster alerts during the CSAT could negatively affect test-takers.

“I have already instructed (the government) to consider extending the exam end time by the amount of time lost, if a test-taker's communication device goes off due to an emergency alert, requiring intervention by a proctor and the test-taker's removal from the testing room,” Lee said.

Lee also set a boundary on the expected response time.

“Unless there are multiple cases of cheating, the necessary measures should take just two to three minutes.”

Lee presented two to three minutes as the normal handling window, not an automatic extension for every candidate. His posts did not say what would happen when several alerts exposed multiple cases of cheating.

What would the time extension cover?

The extension would not apply automatically to every alert. Lee tied it to a specific sequence that begins with a sounding device and ends with removal from the room.

Lee’s disclosed conditions were:

  • An emergency alert causes a test-taker’s communication device to sound.
  • A proctor, meaning a test administrator, must intervene.
  • The test-taker must leave the testing room.
  • The exam receives an extension equal to the time lost during that response.

Those conditions leave operational questions unresolved. Neither Lee’s post nor the government’s announcement said how officials would measure lost time, who would approve an adjustment, or whether every candidate would receive the same extension.

His wording also excludes a merely audible alert from the stated trigger unless it produces both intervention and removal. Without a published measurement and approval process, exam sites could interpret the same interruption differently.

Takeaway: Lee tied added time to a defined security response rather than every alert.

How would alerts help detect AI-enabled cheating?

The government announced its anti-cheating package on Tuesday. Its measures focused on keeping electronic devices out of test rooms and alerting candidates before the exam to reveal devices that had not been surrendered.

The announced measures were:

  • Banning all electronic devices from test rooms.
  • Sending an emergency alert message just before the exam begins.
  • Using that alert to detect mobile phones that candidates had not turned in.

The government described the package as a response to AI-enabled cheating. It did not explain how artificial intelligence would assist cheating, identify any software, or publish evidence about the scale of the problem.

That silence limits independent scrutiny. The announcement establishes a security rationale but provides no baseline for how many violations officials expect or how many alerts they may need.

Together, the measures rest on an unstated enforcement assumption. A complete ban should keep devices out, yet Lee’s compensation clause anticipates a communication device sounding inside a room. The posts do not explain how officials expect that situation to arise.

Lee then proposed extending the alert mechanism to the test itself. During a video meeting about anti-cheating measures, he instructed officials to consider sending random disaster alerts while the exam was underway.

His account places two alert patterns in the same operating plan. The government had announced a pre-exam alert to detect unreturned phones. Lee proposed random alerts during testing as an additional anti-cheating measure.

Lee’s posts did not specify the frequency, sequence, or wording of those random alerts. They also did not say whether the extension would cover every random alert or only those producing the full response he described.

Without those parameters, officials cannot readily distinguish a qualifying security interruption from any other disruption. The proposal therefore creates a second policy question beyond detecting prohibited devices.

Takeaway: The same emergency-alert system serves as a cheating control and a potential source of unequal disruption.

Why was Lee reviewing exam rules from Manila?

Lee chaired the anti-cheating meeting by video link from Manila, where he was visiting to attend the annual ASEAN summit. He disclosed the request after the government had announced the electronic-device ban and the pre-exam emergency alert.

His additional suggestion for random alerts during the test expanded the proposal after those initial measures. The sequence presents the intervention as an iterative policy review rather than a single decision made in advance of the exam.

Neither Lee’s post nor the government’s announcement reported what advice the meeting produced. The available account gives no response from the officials responsible for administering the nationwide exam.

Lee’s wording repeatedly left room for a final decision. He said he had instructed officials to “consider” extending the end time and to consider random alerts, rather than announcing either action as final policy.

The word “consider” does not resolve who will decide or when. The exam was scheduled for Thursday, while the instruction came late Tuesday, leaving the government a narrow window to publish implementation details.

Takeaway: Lee began a policy review, but the available account does not show that either proposal received final approval.

What remains at risk before Thursday?

The immediate policy stakes center on procedural fairness. Lee’s proposal treats time lost during a qualifying security response as something the exam should restore, rather than a cost imposed on the candidate.

Neither Lee’s post nor the government’s announcement explains how officials would prevent unsupported claims or inconsistent treatment. The proposal identifies a fair outcome without specifying the procedure required to reach it reliably.

Lee’s formula addresses one concern by tying added time to an observable sequence. His two-minute-to-three-minute estimate offers a limited planning assumption, except where multiple cases of cheating occur.

Several practical gaps remain. Officials have not said how they would distinguish time spent identifying a device from time spent handling a broader cheating investigation. Nor has Lee said whether an extension applies individually, by session, or across multiple test locations.

The random-alert proposal adds another unresolved issue. More alerts could increase the chance of detecting a prohibited phone, but Lee’s posts provide no detection rate against which to weigh the resulting disruption.

The technical mechanism also remains opaque. The government calls the messages disaster alerts and says one will precede the exam. Lee asks officials to consider more during testing, but neither statement gives a platform, message format, activation method, or detailed notice schedule.

The disclosed measures therefore show the government’s intended sequence without permitting an independent assessment of its safeguards. It has linked AI cheating, emergency alerts, phone detection, possible removals, and time adjustments in one operating plan, but most implementation rules remain unstated.

Takeaway: The proposal states a clear fairness aim but leaves its implementation rules largely unresolved.

The next test is whether Lee’s government can publish a rule before Thursday that identifies qualifying interruptions, calculates added time consistently, and prevents anti-cheating alerts from becoming an uncompensated burden on candidates. If it cannot, the system designed to protect the CSAT could generate the disruption it now seeks to repair.

Source: feed.koreatimes.co.kr

Share this article:

More from James Calloway

James Calloway

Show full bio

News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.

722 articles

Related articles

  1. Han Seong-sook backs loud alerts for AI cheating at CSAT
  2. Korea May Send Unannounced Disaster Alerts During CSAT to Catch AI Cheaters
  3. Korea Will Blast Emergency Alerts Mid-Exam to Catch AI Cheaters
  4. Korea Blocks Wi-Fi at Exam Sites, Bans Pens in AI-Cheating Crackdown
  5. Seoul's Last-Minute CSAT Anti-Cheat Rules Draw Ridicule, Not Trust

« Previous articleNext article »