[AI Brief] Konan Technology Builds Defense Ministry OSINT Analysis Platform

This is an AI brief.

This is a brief service provided by IT Daily.

Konan Technology has won the task of building the 'Open Source Intelligence (OSINT) automatic collection and analysis service' under the 'Defense Ministry AI Applied Product Rapid Commercialization Support' project run by the Institute for Information & Communications Technology Planning & Evaluation (IITP).

Konan Technology said on the 14th that it has begun building an intelligent open source intelligence analysis platform.

The task aims to build a platform that automatically collects, processes and translates open source information in the North Korea and overseas defense and security sectors, and supports intelligence analysis work with AI-based analysis.

OSINT (Open Source Intelligence) is an activity that collects and analyzes materials from legally open information sources such as news, websites and public reports to derive intelligence value. As the volume and types of open information have surged and false and manipulated information has been included, the complexity of the information environment has also increased. Accordingly, the importance of AI-based analysis capabilities that can quickly sift through vast amounts of data and review sources, context and reliability is expanding.

Konan Technology will apply large-scale data collection and processing, natural language processing, search and AI analysis technologies to this project. It also plans to build a specialized information-processing system for defense and security intelligence analysis environments by using multilingual original and translated open source information data, a dictionary of North Korean terminology and military terms, and training data for recognizing people and equipment. Through this, it aims to strengthen timeliness, accuracy and the traceability of analysis evidence.

The new platform covers functions ranging from open source information collection to OCR, multilingual translation, cleansing and classification, source and reliability evaluation, event clustering, signal detection and analysis report generation. The platform's distinguishing feature is that it connects the entire process organically.

The platform is focused on supporting analysts. This will allow analysts to quickly identify information that needs priority review, related events and signs of change.

In line with this trend, the story's theme is the era of 'sovereign AI,' in which companies secure control over their data. Cloudera and Mistral have joined forces.

Cloudera is promoting the creation of a sovereign AI environment that allows companies to use AI while maintaining control over their data. To do this, it combines a hybrid data and AI platform with frontier models and supports public cloud, private cloud, on-premises and air-gapped environments.

Cloudera announced on the 14th that it had signed a strategic partnership with Mistral. The focus of the collaboration is on combining Cloudera's hybrid data and AI platform with Mistral's sovereign AI models, with the goal of supporting enterprises in building, customizing and running AI based on their own data. Cloudera said companies can use AI without taking sensitive information outside and can use AI in their data-resident environments. Supported environments include public cloud, on-premises, edge, sovereign cloud and air-gapped environments.

Cloudera said this approach provides companies that manage large-scale and sensitive data assets with control, choice and flexibility over when, where and how AI operations are run. The company said businesses can run inference in their own environments, customize AI with their own data, and deploy AI applications within existing security and governance frameworks.

The company also said it can manage AI operating costs. Abbas Riki, Cloudera's CBO and head of applied AI, said Cloudera works with Mistral to support the execution and customization of AI where enterprise data resides.

Abbas Riki also said the company helps enterprises maintain control over data, infrastructure and costs. He said the combination of intelligence and control is essential to moving AI from the experimental stage to the practical operational stage. Photo source: ElevenLabs. (ElevenLabs)

ElevenLabs said on the 14th that it has officially appointed Eden Tandowsky, formerly of global fintech payments company Adyen, as CFO. This CFO appointment comes as ElevenLabs transitions from a voice lab to a comprehensive platform for global enterprise AI interaction.

ElevenLabs supports millions of developers, creators and enterprise customers around the world based on voice and audio AI models. Its global team has 800 people, and it supports more than 90 languages.

In addition, ElevenLabs' conversational AI agents number more than 10 million and are being applied across enterprise customer touchpoints such as sales, customer support (CS), recruiting and training. ElevenLabs brought in the new CFO amid this expansion.

New CFO Eden Tandowsky will oversee the expansion of global financial infrastructure and establish a strategic investment framework for next-generation technology research, product development and global market expansion. He also plans to oversee sustainable long-term growth.

ThinkForBL said on the 14th that it is promoting practical AI reliability talent development for university and graduate students, and designed the '2nd Triton' so that participants directly perform reliability-related tasks throughout the AI development process. The goal is to foster talent with practical AI reliability capabilities. The co-organizers are the Korea Network of the Trustworthy AI International Alliance (TRAIN Korea) and ThinkForBL.

Participants will directly carry out impact assessment, risk management, human oversight and agent reliability assurance in actual AI development processes. ThinkForBL announced the opening of the AI reliability hackathon '2nd Triton' on the day.

On the 12th, an orientation for the '2nd Triton' competition was held at Omnibus Park, Seoul St. Mary's Hospital, The Catholic University of Korea. Photo source: ThinkForBL. More than 100 people attended the OJT.

At the orientation and OJT, the difference between AI model performance and AI reliability, the goals of the preliminary and main rounds, submissions and evaluation methods, team composition and role sharing, and how to use the training and provided tools were introduced. The participants' task is to develop a clickbait detection AI model using 'AI Hub' data and SK Telecom's AI model 'A.X'.

In the preliminary round, participants will concretize each team's theme and service direction, analyze users, impacts and risks, and organize reliability activity plans. Teams advancing to the main round will carry out tasks to implement and verify the preliminary plan and complete the final deliverables.

Park Ji-hwan, CEO of ThinkForBL, said AI reliability cannot be formed by mere awareness of standards alone and requires actual response, like first aid. He also explained that the competition is an opportunity to experience the process of fostering practical AI reliability talent with the ability to identify risks based on real projects, judge evidence and change designs when necessary.

Freewheelin participated in the '2025 Edutech Korea Fair.' Related photos were provided by Freewheelin.

At the 'Edutech Korea Fair,' Freewheelin introduced school-tailored AI courseware 'Schoolflat' and university education-focused AI courseware 'FooliCampus' and unveiled its AI courseware.

Freewheelin said it is expanding the scope of application for its AI-based education solutions. The expansion targets are content creation and assessment, and existing functions include worksheet conversion, similar question generation and automatic grading of handwritten descriptive answers. The math-centered service will expand to English and science.

Freewheelin announced on the 14th that it will participate in the '2026 Edutech Korea Fair' and unveil new features of its AI courseware and expanded subject coverage. At the event, it will first reveal AI similar question generation and automatic grading of handwritten descriptive answers. The subjects to be unveiled will also expand from math to English and science.

The new features will be unveiled in the school-tailored AI courseware 'Schoolflat' and the university education-focused AI courseware 'FooliCampus.' These AI courseware products have already supported learning analysis and personalized content recommendations, and now they will also support teacher content creation, teacher grading, and university teaching and learning operations.

Kwon Ki-sung, CEO of Freewheelin, said AI's role in education has expanded beyond learning analysis and recommendations to supporting content creation, evaluation and university teaching and learning operations. He added that the fair will serve as an opportunity to drive continuous advancement of the AI courseware that schools and universities actually need.

Source: IT DAILY · Yang Seung-gap
Original: https://www.itdaily.kr/news/articleView.html?idxno=241598

References

• https://www.itdaily.kr/news/articleView.html?idxno=241598