On October 28, Korean AI assessment solutions company Grep announced that the smart glasses verification feature of its online exam proctoring platform Monito had been adopted by 21 companies and institutions within two months of its August launch. It combines pre-exam multi-angle photo capture with human proctoring to spot smart glasses that could be used for cheating, rather than relying on fully automated AI judgment. Customers include large enterprises in IT, semiconductors, finance, manufacturing, automotive and retail, plus state-owned enterprises, public institutions and broadcasters.

Compiled by VR Gyro. On October 28, Grep, a South Korean AI assessment solutions company, announced that the fine-grained smart glasses verification feature on its online exam proctoring platform Monito had been adopted by 21 companies and institutions within two months of launch. As incidents of cheating in exams using smart glasses keep emerging, such devices are becoming a new challenge for online exam security management. The feature launched in August this year and mainly identifies whether test-takers are wearing smart glasses that could be used for cheating, by combining multi-angle pre-exam photography with human proctoring. Customers adopting the feature so far include large enterprises in IT, semiconductors, finance, manufacturing, automotive, and retail, as well as state-owned enterprises, public institutions, and broadcast organizations. Grep said smart glasses increasingly resemble ordinary eyewear, and with frequent product updates it is hard to accurately identify the device type by appearance alone. Monito therefore does not rely entirely on AI auto-determination; instead it uses AI to assist in capturing device images, giving proctors a basis for verification. Specifically, test-takers wearing glasses must, before the exam begins, use the Monito app to photograph four parts of the glasses: the front, the side, the inner side, and the temple tips. The capture can be done automatically by AI or manually by the test-taker. Proctors can compare the captured images against pre-registered photos and notes, and if anything suspicious arises during the exam, make human judgments informed by the pre-exam records, reducing the adverse impact of AI misjudgment on test-takers. In exam security technology, Grep has previously rolled out a series of features. In November 2025, the company released Monito Plus, an installable secure browser that blocks unauthorized program execution and external device connections; in May 2026, it introduced an AI agent based on large language models (LLMs) that automatically aggregates suspicious behavior during exams and generates reports. Next, Grep plans to add real-time behavior analysis based on vision-language models (VLMs) this November. The feature will identify abnormal behaviors during exams, such as a test-taker switching glasses or repeatedly touching them, and alert the proctors. Grep CEO Lim Sung-soo said the company valued the impact technical misjudgment could have on test-takers when designing Monito, and therefore combines AI image capture with human judgment to give proctors a clear basis for verification. Going forward, the company will further improve the accuracy of online exam proctoring through behavioral context analysis. Source: Biggo

编译/VR陀螺 10月28日，韩国AI测评解决方案企业Grep宣布，旗下在线考试监考平台Monito推出的智能眼镜精细核验功能，在上线两个月内已获得21家企业及机构采用。随着利用智能眼镜违规参加考试的事件不断出现，相关设备正成为在线考试安全管理的新挑战。据了解，该功能于今年8月推出，主要通过考前多角度拍摄与人工监考相结合的方式，识别考生是否佩戴可能用于考试作弊的智能眼镜。目前采用该功能的客户涵盖IT、半导体、金融、制造、汽车、零售等行业的大型企业，以及国有企业、公共机构和广播电视机构。

Grep表示，智能眼镜的外观与普通眼镜越来越接近，加之产品更新频繁，仅凭外观特征难以准确判断设备类型。因此，Monito没有采用完全依赖AI自动判定的方式，而是利用AI辅助采集设备图像，为监考人员提供核验依据。具体而言，佩戴眼镜的考生需要在考试开始前，通过Monito应用拍摄眼镜正面、侧面、内侧及镜腿末端等四个部位。拍摄过程可由AI自动完成，也可由考生手动操作。监考人员能够将采集图像与预先登记的照片及备注进行比对，并在考试过程中出现可疑情况时，结合考前记录进行人工判断，以降低AI误判对考生造成的不利影响。

在考试安全技术方面，Grep此前已陆续推出多项功能。2025年11月，公司发布可安装的安全浏览器Monito Plus，用于阻止未经授权的程序运行及外部设备连接；2026年5月，又引入基于大语言模型（LLM）的AI智能体，自动汇总考试中的可疑行为并生成报告。下一步，Grep计划于今年11月加入基于视觉语言模型（VLM）的实时行为分析功能。该功能将识别考生在考试过程中更换眼镜、反复触碰眼镜等异常行为，并向监考人员发出提醒。

Grep首席执行官Lim Sung-soo表示，公司在设计Monito时便重视技术误判可能对考生造成的影响，因此通过AI图像采集与人工判断相结合的方式，为监考人员提供明确的核验依据。未来，公司还将通过行为上下文分析功能，进一步提高在线考试监考的准确性。来源： Biggo