Why Are Online Exams with Proktor Amanin More Secure?

We have discussed what AI-based proctoring is and why it is becoming increasingly important in the era of digital examinations. We have also explored how Proktor Amanin has evolved through real-world experience with several leading universities and corporations in Indonesia, including ITB, IPB University, Universitas Indonesia, Universitas Terbuka, Telkom University, Telkom ACI, Schoters, Adeline Foundation, and other institutions.

Now, we come to the question most frequently asked by examination organizers: Are online exams conducted with Proktor Amanin really secure?

This is a legitimate question. A case once went viral on social media involving a university entrance examination candidate who was declared accepted, despite allegations that they had used AI assistance to answer the exam questions. The method was simple: the exam questions were photographed and uploaded to an AI application commonly used by students to complete their homework. Cases like this have prompted some universities to return to on-campus examinations or computer-based testing (CBT), even though these approaches are significantly more expensive for both universities and participants.

The problem, however, is not online examinations themselves. The problem lies in how they are monitored. Many online examinations are still supervised using video conferencing applications that were never designed for examination proctoring. There is no AI continuously monitoring participants' faces, no system monitoring their screens, and technical assistance is provided through WhatsApp or Telegram.

Security Is Not About One Feature, but Multiple Layers

The short answer is that online exams with Proktor Amanin are more secure not because of one sophisticated feature, but because security is built through multiple layers. Cheating in online examinations takes many forms. There may be impersonators taking the exam on behalf of registered candidates. Other people may provide assistance from beside or behind the participants. Candidates may open search engines or AI applications, take screenshots, photograph questions using their smartphones, or ask AI applications questions through voice commands.

Because cheating methods vary, security measures must also be comprehensive. One layer may be bypassed, but penetrating all layers simultaneously is considerably more difficult.

Six Layers of Proktor Amanin Security

Layer 1: Ensuring the Right Person Takes the Exam — Initial Identity Verification

Security begins before the first examination question is opened. When Proktor Amanin was initially developed, competing products used by universities still relied on manual participant verification. Candidates had to face the camera while showing their identification cards, which examination officers then checked one by one. Amanin chose a different approach: AI-based facial recognition, leveraging the company's earlier research and development in facial recognition technology.

The system compares each participant's face with the photograph submitted during registration. If the faces do not match, the participant cannot access the examination questions. This is the first line of defense against impersonation. Our customers' experience has shown that this approach is more efficient and carries a lower risk of errors compared with manual verification.

Layer 2: Continuous Monitoring — Ten AI Monitoring Parameters

Passing the initial identity verification is not enough. What if an impersonator takes over after the examination has started, replacing the registered participant or sitting beside them?

Together with our customers, Amanin initially formulated five AI monitoring parameters. Based on real-world experience, these parameters have now evolved into ten: eight monitored through the laptop camera and screen, and two through the smartphone camera (Mobile Pairing, discussed in Layer 3). The ten monitoring parameters are:

1. Face Out of Frame — The participant leaves the camera's field of view or repeatedly looks around.
2. Multiple Faces Detected — Another person is present and potentially providing assistance.
3. Face Does Not Match Registered Identity — The participant has been replaced by an impersonator.
4. New Tab or Browser Opened — The participant attempts to search for answers from external sources, including AI applications.
5. Print Screen or Screenshot Detected — An attempt to capture and distribute examination questions to obtain outside assistance.
6. Copy/Paste Activity Detected — Copying examination questions externally or pasting answers from another source.
7. Excessively Loud Audio — Conversations or assistance from other people around the participant.
8. Object Detected — An object that should not be visible within the laptop camera's field of view.
9. Object Detected (Smartphone) — An object that should not be visible within the smartphone camera's field of view.
10. Incorrect Body Position (Smartphone) — The participant's body position does not comply with the required positioning, preventing proper monitoring through the smartphone camera.

All ten parameters operate continuously throughout the examination. Every violation is recorded as a warning with its own corresponding code, allowing proctors to rank participants based on the total number of warnings or specific types of violations. Participants' identities and behavior are therefore not checked just once. They are monitored from beginning to end.

Layer 3: Seeing and Hearing Beyond the Screen — Audio Detection and Mobile Pairing

A laptop camera can only see what is directly in front of it. Yet cheating often happens precisely outside that field of view. That is why Proktor Amanin incorporates audio detection.

Its value became immediately apparent during a university entrance examination. A proctor heard a fairly loud conversation coming from the direction of one participant. Upon listening more carefully, the proctor discovered several impersonators sitting behind the participant. The proctor warned the participant through private chat and, if necessary, through a voice call. After receiving three warnings without responding appropriately, the participant was disqualified.

Cases like this led to the development of Mobile Pairing. By scanning a QR code, a participant's smartphone becomes a second camera connected to the system. The laptop camera monitors the 180-degree area in front of the participant, while the smartphone camera monitors the other 180 degrees. Impersonators hiding behind participants, who previously could not be seen by the laptop camera, can now be detected.

Layer 4: Monitoring What Happens on the Screen — Screen Lockdown and Activity Monitoring

In the era of generative AI, the biggest threats are often invisible to cameras because they take place directly on the participant's screen. Cheating methods vary, including opening ChatGPT or StudyX.ai, using OCR to read examination questions, and mirroring the screen to another device. Proktor Amanin addresses these vulnerabilities from two directions.

First, prevention: the system can lock the screen, preventing participants from opening other applications or switching browser tabs. Second, monitoring and recording: browser and window activities, copy/paste actions, screenshots, and even screen recordings are captured and logged with their respective timestamps.

One of our customers' experiences demonstrates the value of this security layer, even for examinations conducted in physical classrooms. Previously, some students had accessed prohibited applications during on-campus examinations without being noticed by the examination assistants. During one examination involving 220 students across five rooms monitored by Proktor Amanin, the system recorded two students opening other applications: one opened WhatsApp and immediately closed it, while another opened Google Translate. Was this a violation? The decision rests with the lecturer. The system's responsibility is to monitor activities and provide evidence.

Layer 5: AI Observes, Humans Decide

AI excels at consistency. It does not get tired after two hours and can monitor thousands of participants simultaneously, including their facial expressions, movements outside the camera frame, unusual sounds, and signs of restlessness. Every finding appears as a flag on the proctor's dashboard, complete with timestamps and recordings. However, an indication is not a verdict. A participant who looks away may simply be thinking. That is why, in Proktor Amanin, the final decision always remains in human hands.

Purely manual monitoring also has its own weaknesses: different proctors may apply different standards. With AI acting as the initial screening mechanism, proctors no longer need to stare at dozens of tiny faces on a screen. Instead, they can focus their attention on participants who genuinely exhibit suspicious behavior. This approach enables one proctor to supervise approximately 45 participants at Universitas Indonesia, 50 at ITB, and 65–75 at IPB University.

Layer 6: Evidence That Can Be Reviewed — Post-Proctoring

Post-proctoring has been part of Proktor Amanin since its early development. Each participant has an individual violation report, accompanied by camera recordings, screen recordings, audio recordings, activity logs, and a list of AI-generated flags.

For examination organizers, this comprehensive evidence is crucial for academic decisions, internal audits, and process transparency. These records enable data-driven evaluations rather than decisions based on assumptions. Even participants suspected of misconduct benefit from a fairer process because decisions are based on verifiable evidence.

Supporting All Security Layers: An Integrated Help Desk

There is one vulnerability that is often overlooked: what happens when participants experience technical problems? If the only way to request assistance is through WhatsApp or Telegram, participants must leave the examination environment. That is precisely where another security vulnerability can emerge.

Proktor Amanin's Help Desk is integrated directly into the application. Participants can chat or speak directly with support officers, who can escalate issues to IT Support when necessary. All reported problems are recorded by the system and become valuable input for further system improvements. Monitoring continues, technical issues are addressed, and participants do not need to open other applications. This is what we call the experience of an online examination with the convenience and discipline of an offline examination.

More Secure Compared to What?

The term "more secure" needs a point of comparison. The following table compares Proktor Amanin with two commonly used examination methods: online examinations conducted through video conferencing platforms such as Zoom, and conventional on-site examinations in physical examination rooms.

This comparison shows that conventional on-site examinations also have limitations. Several real cases demonstrate this: supervisors in examination rooms can see the students, but they cannot simultaneously monitor 220 individual laptop screens. In many respects, online examinations with layered monitoring can actually be more rigorous than examinations conducted in physical classrooms.

Proven in Real-World Implementations, Not Just Demonstrations

The security of a system can only truly be tested under real-world conditions, with thousands of participants, different types of devices, and internet connections that are not always stable.

• ITB has conducted fully remote entrance examinations since 2023, including sessions involving more than 10,000 participants at once, without significant problems.
• IPB University previously supervised examinations using Zoom. After comparing several methods and conducting trials, IPB selected Proktor Amanin for its 2025 Independent Admission Examination and subsequent examinations because Zoom was no longer adequate for large-scale examinations and increasingly sophisticated cheating methods.
• Universitas Indonesia selected Proktor Amanin for SIMAK UI 2025 following technical discussions, demonstrations, and participant simulations. The university has continued using Proktor Amanin to this day. UI reported that approximately 80% of its objectives had been achieved and plans to use the system again for postgraduate entrance examinations. The examination involved more than 23,000 participants.
• Universitas Terbuka began adopting Proktor Amanin for its 2026 semester examinations. Proktor Amanin was selected through a highly competitive tender process involving both domestic and international vendors with extensive experience in the proctoring industry.

The experiences of these customers provide an important lesson: security is not determined by technology alone. The readiness of the organizing committee, internal procedures, training for proctors and participants, and coordination between academic and IT teams are equally important. These institutions conducted simulations and provided training for participants before the examinations. Technology is a tool; success comes from synergy.

Being Honest About Its Limitations

Does this mean cheating can be reduced to zero percent? No examination monitoring system, whether online or offline, can guarantee that. As the saying goes, thieves are always one step ahead of the police. That is why Proktor Amanin continuously introduces new features to address evolving cheating methods, such as Mobile Pairing, which was developed in response to an actual case involving impersonators hiding behind participants.

The objective of security is not perfection. It is to make cheating extremely difficult, highly risky, and very likely to be detected. When participants know that their identities are verified, their faces are continuously monitored, their surroundings are visible from two different angles, their screen activities are recorded, and all evidence can be reviewed, their motivation to cheat is reduced. Effective monitoring does not merely detect cheating. It prevents it from happening in the first place.

Secure for Examination Organizers, Fair for Participants

So, why are online examinations with Proktor Amanin more secure? Because security is built through multiple layers: initial identity verification, ten AI monitoring parameters throughout the examination, audio detection and Mobile Pairing, screen lockdown and activity monitoring, collaboration between AI and human proctors, and post-proctoring evidence. All these layers are supported by an integrated Help Desk.

However, good security does not only protect examination organizers. It also protects honest participants. Students who have studied hard deserve to know that their results will not be undermined by impersonators or AI applications. Participants who are suspected of misconduct also deserve a fair evaluation process based on evidence, not mere suspicion.

There is an even bigger issue at stake. Allowing students to cheat is essentially training them to be dishonest when they eventually enter the professional world. The risk is too great for our nation if its future leaders are allowed to pass selection processes through dishonest means.

That is the true meaning of security: not merely making cheating difficult, but ensuring that examination results can be trusted by everyone. Because ultimately, the safest online examination is not the one that catches the most offenders. It is the one that makes everyone confident in its results.

Want to see how these layers of security work for examinations at your institution? Contact us via WhatsApp: +6287885110303.

Back to home