- HOME
- IT·SCIENCE
-
'My tastes are gone — only the AI's remain': the paradox of ChatGPT recommendations
- UNIST develops EchoTrace to track risks in AI recommendation systems - Vicious cycle of reliance on past recommendations amplifies bias with each iteration The movies, books and restaurants that generative AI systems such as ChatGPT recommend may be shaped less by a user's actual preferences than by what the AI itself recommended in the past, according to new research. The Ulsan National Institute of Science and Technology (UNIST) announced Sunday that a research team led by Professor Lee Yeon-chang of its AI Graduate School has developed EchoTrace, a diagnostic technology that tracks errors and bias in large language model (LLM)-based recommendation systems. Personalized recommendation services powered by generative AI have been spreading rapidly across online shopping platforms and video streaming services. These systems analyze user preferences to suggest products, or generate synthetic user profiles to improve recommendation accuracy. Problems can arise, however, when AI-generated recommendations are fed back into the system as training data. The products an AI previously recommended carry more weight in shaping the next round of suggestions than a user's actual preferences do, creating a self-reinforcing distortion. To examine this problem, the research team analyzed recommendation systems for movies, books and restaurants. When AI directly generated recommendation lists, the share of items presented that did not exist in the eligible pool reached 4.07 percent for movies, 7.40 percent for books and 13.14 percent for restaurants. This is the so-called "hallucination" phenomenon, in which AI presents nonexistent or ineligible items as if they were normal recommendations. Serious errors also emerged when the AI tried to infer user information. In 93.16 percent of cases, the occupation the AI estimated based on a user's movie-viewing history did not match the actual job — and in some instances the system invented occupations entirely, such as "film critic," a category absent from standard job classifications. A concentration of recommendations on certain popular products was also observed. When AI-generated training data was used, the system tended to over-recommend popular films, specific genres or particular business categories compared with recommendations derived from actual user records. The deeper problem is that these errors do not disappear even after repeated retraining. When the research team ran the retraining process five times using AI-generated recommendations, the initial biases and errors persisted. In some systems, the skew toward popular products grew worse. Particularly in systems where AI directly recommends films, the AI's own previous recommendations were the single most influential factor in generating the next round of suggestions for between 86 and 92 out of every 100 users. A feedback loop forms in which AI-generated recommendations are recycled as training data — treated as records of actual user choices — and then used to produce new recommendations. The more this cycle repeats, the less likely it becomes that a user's genuine preferences are reflected. EchoTrace is designed to analyze these risks systematically. While conventional evaluations of recommendation systems focus on accuracy at a single point in time, EchoTrace tracks how AI-generated information and recommendations change as they pass through repeated rounds of retraining. The team analyzed 116 papers published at international academic conferences to classify the roles LLMs play in recommendation systems into five categories, then verified the errors and biases arising at the stages of information generation, final recommendation and iterative retraining for three of those roles. "This research has laid the groundwork for systematically diagnosing how bias and hallucination in LLM-based recommendation systems accumulate and amplify through repeated recommendation cycles," Lee said. "It can be used going forward to design and evaluate recommendation systems that are safer and more trustworthy."
Nov. 8, 2026
-
AO3 fan fiction site down for over a week after cyberattack
User database may have been compromised; platform won Hugo Award in 2019 One of the world's most popular fan fiction sites has been inaccessible for more than a week following a cyberattack. AFP reported Saturday that Archive of Our Own (AO3), an online fan fiction platform, has been down since Oct. 29 — nine consecutive days as of Saturday. The Organization for Transformative Works (OTW), which operates AO3, attributed the outage to an external cyberattack and said the database storing users' email addresses and passwords may have been stolen. However, OTW said no fan fiction content posted on the platform had been lost. AO3 launched in 2008 and has more than 11 million registered users and approximately 18 million posted works. The platform won the Hugo Award, one of the world's most prestigious science fiction literary prizes, in 2019. Meanwhile, the US government recently issued a statement warning of continued cyberattack threats targeting industrial refrigeration and freezer systems. On Monday, hackers also struck supermarkets in Sweden, Belgium and other European countries, forcing the disposal of large quantities of food.
Nov. 7, 2026
-
Naver, Google race to put user-generated content at the heart of search
Naver and Google are racing to put user-generated content at the center of their search services. Naver has rolled out a new search feature that groups user experiences and reviews by topic, while Google has unveiled a program that lets external online communities feed their latest posts directly into its search index. As competition in AI-powered search intensifies, content rooted in real user experience is emerging as a key differentiator. According to the IT industry, Naver introduced an "experience UGC block" to its search service on Thursday, organizing posts around shared topics of interest rather than simply listing results for a given keyword. When a user searches "Fukuoka travel," for example, subtopics such as "Fukuoka day trips to nearby areas" appear in the results, accompanied by related posts from Naver blogs and cafes. The feature lets users compare experiences and reviews relevant to their interests without sifting through individual posts one by one. Rather than displaying full post text, the feature excerpts only the portions relevant to each subtopic. Comments on cafe posts are also shown, and short tags summarize the experiences described in each post so users can grasp at a glance what the author actually did or encountered. Naver is leveraging its own UGC ecosystem as a core competitive asset — connecting accumulated data from its blogs, cafes and Knowledge iN service directly to its AI search offering, rather than relying solely on crawling external websites. In-house UGC already accounts for a substantial share of Naver's AI search output. According to Naver, as of January this year, its own UGC made up 70 percent of the content cited in AI briefings. Naver announced in May that it plans to invest 1 trillion won ($745 million) in its content ecosystem over the next five years. It has also launched a "Naver Mate" program that selects top creators frequently cited in AI briefings and pays them activity fees — a strategy aimed at encouraging content production and channeling it back into search competitiveness in a virtuous cycle. Naver is also pushing back against unauthorized data collection by outside AI companies. Since last year, it has blocked external AI crawlers from accessing its UGC by default. Google is similarly focused on enhancing UGC-driven search, though its approach differs from Naver's in one key respect: rather than drawing on a proprietary content ecosystem, it is leaning heavily on partnerships with external platforms. On Oct. 8, Google unveiled its "UGC Fresh Data Program," which allows online communities and social media platforms to transmit their latest posts and user-engagement data directly to Google. Unlike the conventional method in which Google's search crawlers collect information from web pages, the program gives approved platforms a dedicated data-transmission channel to send user-response data straight to Google — a move aimed at reflecting fast-changing community content in search results more quickly. This is not Google's first push to secure UGC data. The company previously expanded its data partnership with Reddit, the US online community platform, in 2024, gaining access to Reddit's API to receive real-time posts and structured data.
Nov. 7, 2026
-
Tesla Cybertruck to be discontinued after dismal sales — just 20,000 units a year despite 1 million pre-orders
The Cybertruck's noisy debut is fading to silence. Tesla's Cybertruck, which generated strong buzz when it was unveiled, is being discontinued after years of sluggish sales. Tesla CEO Elon Musk had boasted that the truck racked up more than 1 million pre-orders at launch, but actual annual sales turned out to be just over 20,000 units. First unveiled in 2019, the Cybertruck drew wide attention for its distinctive angular design. But poor ride quality, repeated recalls, persistent quality issues and a steep price tag were cited as the main reasons for its weak sales. In South Korea, the vehicle was priced at 160 million won ($119,000). Musk explained on X, formerly Twitter, on Friday (local time) that Tesla plans to "wind down Cybertruck production at the Texas Gigafactory around the middle of next year" to free up space for manufacturing the Cybercab. Foreign media outlets said the announcement effectively signals the end of the Cybertruck, as the Austin, Texas, Gigafactory is the only facility capable of manufacturing the truck's specialized components. Responding to a question from an X user, Musk called the Cybertruck "the best vehicle Tesla has ever made" and described the decision as "the end of an era." According to automotive data platform Kelley Blue Book, US sales of the Cybertruck totaled 20,237 units last year. Cox Automotive, an auto market research firm, found that sales fell 28 percent year-on-year in the first nine months of this year. US registrations in October were limited to 1,914 units, according to Motor Intelligence data. A significant portion of those sales went to SpaceX, the space company Musk leads, which was one of the Cybertruck's largest customers last year. According to SpaceX's initial public offering filing, the company purchased $131 million worth of Cybertrucks from Tesla in 2025 at list price. Tesla has been shifting its strategic focus toward robotics and autonomous driving technology. The Cybercab — which features butterfly doors that open upward and lacks a steering wheel, pedals and mirrors — is one of the key bets Musk is placing on transforming Tesla from a simple automaker into an AI and robotics company. Reuters noted, however, that the Cybercab is unlikely to offset the production volume lost from winding down the Cybertruck.
Nov. 7, 2026
-
CJ ENM's film and drama unit swings to profit after years of losses
CJ ENM's film and drama division swung to a profit of 12.9 billion won ($9.61 million) in the third quarter of this year, ending a prolonged streak of losses. The company plans to sustain the momentum in the fourth quarter with upcoming titles including "100 Days of Lies" and "My Guilty Person." The turnaround comes after years of stumbles that seemed at odds with CJ ENM's reputation as one of South Korea's premier entertainment companies. Its share price, which once approached 300,000 won, fell below 30,000 won in July. According to industry sources, CJ ENM's film and drama division posted a cumulative operating profit of 12.9 billion won for the third quarter of this year, on a preliminary basis. The result marks a reversal after consecutive annual losses — 97.5 billion won in 2023, 41.3 billion won in 2024 and 1 billion won last year — and a further 2.6 billion won loss in the first half of this year. CJ ENM is widely regarded as one of South Korea's leading content companies, yet its share price has been in steep decline since the late 2010s. After approaching 300,000 won in 2018, the stock slid sharply as the company failed to produce a major hit following "Parasite." It bottomed out at 28,100 won on July 1. The box-office failure of "Tazza: The Song of Belzebub," a film that had generated considerable internal excitement, dealt a further blow to the company's film business. At the time, shareholders vented their frustration online, with comments such as "just cut your losses and get out" and "there's no hope." The third quarter of this year has now given the content division a chance at a comeback. CJ ENM attributed the improved profitability to supplying the Netflix original "Slowly, Intensely" and expanding pre-sales of titles including "Four Hands" and "See You at Work Tomorrow" to global OTT platforms. In July, the company also signed a content partnership with Amazon Prime Video India, broadening its global distribution network. CJ ENM said it intends to keep the content division in the black through the fourth quarter, pinning its hopes on "100 Days of Lies," "My Guilty Person" and "Paradise Lost." The company said tvN's "100 Days of Lies" has been performing well in both viewership ratings and buzz, while "My Guilty Person" has held the No. 1 spot in its time slot, demonstrating strong commercial results. CJ ENM added that it plans to sustain the momentum with director Yeon Sang-ho's film "Paradise Lost," which opened in October, followed by premium content including "Gift" in December. Meanwhile, CJ ENM recently filed a preliminary disclosure showing cumulative sales of around 3.72 trillion won and operating profit of around 57 billion won through the third quarter of this year, on a consolidated basis.
Nov. 6, 2026
-
'Nobody cared, nobody believed me': S. Korea ranks 2nd among 10 countries in AI deepfake sex crime victimization [Huam-dong Research Lab]
Survey of 7,230 adults finds 7% victimized by AI-generated sexual images Women reported higher rates of depression and fear; men more likely to say they were fine or found it funny More than half of perpetrators were people victims knew only online About 7 in 100 adults in the United States, the United Kingdom and Australia have been victimized by AI-generated fake sexual images, according to new research. So-called deepfake sex crimes — in which AI is used to superimpose someone's face onto pornographic material — are growing rapidly. One analysis found that 96 percent of deepfake videos posted online in 2019 were non-consensual sexual images. Sites offering "undressing apps" — tools that generate images making a person appear unclothed from a single photo — drew an average of about 18.5 million visitors per month in the first half of last year across 85 platforms, and were estimated to earn up to $36 million annually. Yet most research to date has focused on the victimization of celebrities and public figures, leaving little known about how ordinary people are affected and what harm they suffer. A team led by Nicola Henry, a professor at RMIT University in Australia, and researchers from Google recently published findings in the International Journal of Human-Computer Interaction based on a survey of 7,230 adults in the US, UK and Australia on AI deepfake sex crime victimization. 15.1% victimized by basic image manipulation; 7% by AI-generated fakes The research team surveyed adults in the US, UK and Australia in July and August 2024, asking whether anyone had ever created or threatened to distribute non-consensual fake sexual images of them since they turned 18. When basic image manipulation tools such as Photoshop were included, 15.1 percent of respondents said they had experienced such victimization. Of those, 7.0 percent reported being victimized specifically by AI-generated images. Among respondents, 4.3 percent said someone had used AI to create sexual images of them, 4.4 percent said such images had been distributed, and 3.6 percent said they had been threatened with distribution. In an earlier survey the same research team conducted across 10 countries, Australia recorded the highest rate of AI sexual image victimization at 3.7 percent, with South Korea second at 3.1 percent. Social media posts were the most common channel through which harmful images spread, accounting for 39.4 percent of cases. Private messages accounted for 32.9 percent, images shown to others on a screen for 30.4 percent, and group chats for 28.9 percent. Victimization rates were higher among those under 35, LGBTQ+ individuals, people of color and people with disabilities. Men reported victimization at more than twice the rate of women. However, the research team said this may be because male victims are more likely to be aware when they are targeted through sexual extortion or pranks, while images created and circulated covertly may go entirely undetected by victims. Women reported depression and fear; men more likely to say they were fine or found it funny According to the research team, 90.3 percent of victims experienced negative emotions such as anger or shame. A woman in the UK aged 25 to 34 said she "felt humiliated watching the video" and "couldn't pull myself together for a long time." Another woman in the UK aged 18 to 24 said it was "degrading" and that "it happened in an instant, but what followed was unbearably long." Female victims were significantly more likely than male victims to report depression (42.6 percent vs. 25.7 percent), fear for their physical safety (42.4 percent vs. 18.1 percent) and sadness (32.7 percent vs. 10.7 percent). Women were also more than twice as likely as men to say their mental health had deteriorated — 35.5 percent compared with 16.0 percent. Among male respondents, 11.4 percent said they were fine, and 12.3 percent said they found it funny. More than half of perpetrators — 55.6 percent — were people the victim knew only online. Those the victim had met in person accounted for 29.7 percent, while 14.6 percent of victims did not know who was responsible. Victims most commonly responded by blocking the perpetrator (69.2 percent) or reporting them to the platform (56.8 percent). More than half also gathered evidence (55.0 percent) or requested that the images be removed from the site (52.6 percent). Meanwhile, 33.0 percent of victims were unaware of StopNCII, an online tool designed to help prevent the spread of non-consensual intimate images. About 1 in 10 victims (12.0 percent) neither told anyone nor filed a report. A woman in Australia aged 35 to 44 said she "was too embarrassed." A woman in the US in the same age group said she "didn't know who to report it to or what good it would do." A woman in the UK aged 18 to 24 said "nobody cared, and nobody believed me or listened to me." The research team said follow-up studies are needed to examine victimization among those under 18 and in countries outside the Western world. Reference DOI: 10.1080/10447318.2026.2723677 Henry, N., Umbach, R., Shelby, R., & Gonzalez-Pons, K. (2026). "That Experience Still Haunts Me to This Day": A Multi-Region Survey on AI-Generated Image-Based Sexual Abuse Victimization. International Journal of Human-Computer Interaction, 1-26.
Nov. 6, 2026
-
Hundreds of microplastic particles found in cup ramyun, study finds
Too much trouble to boil ramyun on the stove? Ramyun is South Korea's undisputed national food. The country consumes about 4 billion individual servings a year — roughly one per person every four to five days. Cup ramyun takes that convenience a step further: just add hot water and wait, no pot or stove required. But a hidden problem lurking behind that convenience has now come to light. Hundreds of microplastic particles are being released into the water each time hot water is poured into a cup ramyun container. The culprit is the cup itself — specifically, the thin plastic coating applied to the inside of the container to prevent leaks. That is why experts now advise choosing packaged ramyun cooked in a pot over cup ramyun. Researchers at China Agricultural University recently published findings on microplastic release from cup ramyun containers in the Journal of Food Composition and Analysis, according to the climate and environmental science community Friday. The team examined 10 cup ramyun brands from four countries — South Korea, China, Japan and Thailand — obtaining two products per brand for a total of 20 samples. The experiment was straightforward. Researchers poured 500 milliliters of water heated to 95 degrees Celsius into each cup, covered it with aluminum foil and waited 15 minutes. They then removed the foil, left the cups to stand for another 20 minutes and analyzed the microparticles present in the water. Microplastics and other microparticles were detected in all 20 samples. According to the paper's abstract, the average number of microparticles found per cup reached 318 — hundreds of invisible particles released simply by pouring in hot water. The substances detected were varied. Researchers identified multiple types of plastic, including polyethylene (PE), a polyethylene-polypropylene (PE-PP) composite and polyester. Where are these microplastics coming from? The researchers focused on the container itself. The inner surface of many cup ramyun containers is coated with a thin plastic film to prevent water from seeping through, and the team suggested that microparticles are likely breaking off from this plastic coating layer. This is not a new concern. In April last year, domestic test results showing microplastics detected in cup ramyun containers were made public in South Korea. Analysis by KOTITI Testing and Research Institute, a nationally accredited testing body, released by the office of People Power Party lawmaker Im Yi-ja, found that pouring water at 100 degrees Celsius into a styrofoam cup ramyun container sold under a domestic brand and leaving it for 30 minutes produced 17 microplastic particles. Cup ramyun containers sold in South Korea fall broadly into two categories: styrofoam-based and paper. Polystyrene styrofoam containers were once the dominant choice, prized for their insulating properties and low cost. As concerns grew over the environmental impact and safety of single-use plastics, however, some products switched to paper containers. Styrofoam has not disappeared entirely — polystyrene containers are still used in cup ramyun sold in South Korea today. The problem is that paper containers are not entirely free of plastic either. They rely on an inner plastic coating to keep the paper from absorbing liquid, meaning the "paper cup" label does not mean plastic-free. How harmful is it to ingest microplastics from cup ramyun? No clear conclusion has been reached. Further research is needed to determine how much exposure over how long a period begins to affect human health. Concerns about the potential adverse effects of microplastics on the human body have, however, been raised consistently. Animal studies have documented changes in sex hormone levels and negative effects on reproductive function in subjects exposed to microplastics. Some research has also observed increased inflammatory responses and oxidative stress in the body. South Koreans are particularly heavy ramyun consumers. According to the World Instant Noodles Association, domestic ramyun consumption last year reached about 4.06 billion servings, ranking ninth in the world. Divided by population, that works out to roughly 79 servings per person per year — about once every four to five days — placing South Korea second in per capita consumption globally, behind only Vietnam. The potential risks of microplastic exposure are therefore difficult to dismiss, and the government has launched investigations to assess the actual situation. Under the government's Sixth Basic Plan for Food Safety Management, drawn up late last year, authorities plan to conduct 300 microplastic surveys this year targeting food utensils, containers and packaging materials. From 2029, surveys on microplastic exposure through food consumption are also set to begin. In the meantime, consumers looking to reduce unnecessary exposure can consider cooking ramyun in a separate pot rather than using the disposable cup. Experts have also said that minimizing the use of disposable containers through alternative methods is important. The researchers themselves stressed the point. "Disposable cup ramyun containers are a major source of microplastic release," they said. "To reduce microplastic intake, it is important to cut back on cup ramyun consumption or avoid drinking the broth."
Nov. 6, 2026
-
A quarter of the standard dose still cuts liver enzyme levels in fatty liver patients, study finds
200 IU of vitamin E cuts ALT levels by 38% in fatty liver patients, trial of 200 finds 16 combined clinical trials also show liver enzyme improvement, but researchers warn against high doses Effects on liver fibrosis and long-term outcomes remain unknown 'My checkup showed fatty liver — is there anything I can take?' Fatty liver disease is the most common chronic liver condition. It is estimated to affect between one in five and one in three adults worldwide. Even without heavy drinking, fat can accumulate in the liver in people with obesity, diabetes or high cholesterol. In some cases, the condition progresses to inflammation and fibrosis, a hardening of liver tissue. The standard approach to treatment is dietary changes and exercise. New drugs for fatty liver inflammation have come to market recently, but concerns have been raised that their high cost makes long-term use difficult. One supplement that has long been discussed as an alternative is vitamin E. As an antioxidant, it is thought to reduce damage to liver cells. Most research confirming its effectiveness, however, has used high doses of 800 IU or more per day, raising questions about the safety of prolonged use. This year, several notable findings from international research have addressed that dosage question. Even a quarter of the standard dose lowers liver enzyme levels A research team from nine hospitals, including the Cleveland Clinic, published results in June in the international journal Hepatology from a clinical trial in which 200 fatty liver patients were given vitamin E at varying doses. Participants had high liver fat content and elevated ALT levels of 60 or above. Their average age was 47. The research team divided participants into four groups, giving them natural vitamin E at 200 IU, 400 IU or 800 IU per day, or a placebo, for 24 weeks. ALT levels fell by 38 percent in those who took 200 IU, and by 36 percent in those who took 400 IU and 800 IU respectively. Those who took the placebo saw a 12 percent reduction. Those taking the lowest dose of vitamin E, in other words, saw liver enzyme levels drop by roughly the same amount as those taking much higher doses. More participants in the vitamin E groups also saw their ALT return to the normal range. Among those who took 200 IU, 49 percent reached normal levels, as did 50 percent of those who took 800 IU, compared with just 22 percent in the placebo group. When participants were monitored for 24 weeks after stopping the supplement, those who had taken vitamin E maintained ALT levels lower than at the start of the trial. Side effects were no different from those in the placebo group. In the paper, the research team said the findings "challenge the conventional wisdom that doses above 800 IU are necessary for a therapeutic effect." However, the research team said it was too early to conclude from these results alone that 200 IU is as effective as 800 IU. Benefits confirmed, but high doses should be avoided The finding that vitamin E lowers liver enzyme levels has been confirmed by other research as well. A research team from Shahid Beheshti University of Medical Sciences in Iran and other institutions published results in June in the international journal Nutrition & Metabolism, combining data from 16 clinical trials involving fatty liver patients. Participants who took vitamin E saw significant reductions across all four liver enzyme markers measured, including ALT. The effect was more pronounced in those who took it for more than 12 weeks and in those with obesity. The research team also warned, citing existing analyses, that high-dose vitamin E of 400 IU or more per day should be avoided because it may increase the risk of death. The same team had earlier published a separate analysis in May in the international journal Lipids, pooling data from 12 clinical trials involving 1,166 participants, finding that vitamin E lowered triglyceride levels in fatty liver patients. The reduction was greater when taken at relatively lower doses over a longer period. Improved liver enzyme levels do not necessarily mean fatty liver disease has been cured. The Iranian research team said the available evidence makes it difficult to determine what effect vitamin E has on the progression of liver fibrosis or on long-term liver health. Natural vs. synthetic form also matters The vitamin E used in the US clinical trial was the natural form, d-alpha-tocopherol. The US research team noted that many products on the market contain the synthetic form, dl-alpha-tocopherol, which may be metabolized differently in the body, and said it remains unknown whether the synthetic form would produce the same effects. The US research team excluded from the trial people taking blood thinners, those at risk of bleeding, and men at high risk of prostate cancer. Both research teams also said vitamin E cannot replace dietary changes and exercise, and should be used only as a supplementary measure alongside them. Reference papers DOI: 10.1097/HEP.0000000000001797 Dasarathy S, Mitchell EP, Wilson LA, et al. Vitamin E dosing study (VEDS) in patients with metabolic dysfunction-associated steatotic liver disease with elevated aminotransferases: a multicenter, randomized, placebo-controlled trial. Hepatology. Published online June 9, 2026. DOI: 10.1186/s12986-026-01158-5 Karimi M, Abidi M, Gholamian R, et al. Vitamin E administration reduces liver enzyme levels in patients with metabolic dysfunction-associated steatotic liver disease (MASLD): evidence from a systematic review and meta-analysis. Nutr Metab (Lond). 2026;23:105. DOI: 10.1002/lipd.70042 Karimi M, Abidi M, Aghaei A, et al. Efficacy of vitamin E on lipid profile in patients with metabolic dysfunction-associated steatotic liver disease (MASLD): a systematic review and meta-analysis. Lipids. Published online May 6, 2026.
Nov. 6, 2026
-
Fly drones, pilot robots: science museum to host advanced technology showcase
National Science Museum to run 'Advanced Technology Expedition' program Dec. 5-6 Children will get to fly drones and come face to face with humanoid robots and AI robots at a special science experience event set to open next month. The National Science Museum announced Friday that it will run the winter season of its "2026 Advanced Technology Expedition" program Dec. 5-6 at the special exhibition hall on the third floor of the Future Technology Building at the Daejeon museum. The event is designed to give children who rarely encounter cutting-edge science and technology a chance to experience a range of future technologies firsthand. Participants will be able to see and operate drones, autonomous driving robots, humanoid robots and AI robots, learning how each technology works through hands-on interaction. The Advanced Technology Expedition is one of the National Science Museum's flagship science outreach programs, created to narrow the gap in access to science culture across different regions. Since last year, the program has traveled directly to elementary schools in remote and underserved areas nationwide, showcasing advanced technologies including drones, quadruped robots and humanoid robots. This year, the museum expanded beyond its outreach model by inviting children from child welfare facilities to visit the museum itself — broadening the experience so that children can explore the museum's full range of facilities and content, not just receive a technology demonstration. After the event, the National Science Museum plans to connect participants with online science learning content and a metaverse science museum to help sustain children's interest in science. On the first day, Dec. 5, the program will be open to elementary school students from child welfare facilities, including community child centers, care centers and family centers. On Dec. 6, the general public may also participate. The program will run three times daily — at 10 a.m., 1 p.m. and 3:30 p.m. Each session is capped at 100 participants, and admission is free. Elementary school students from child welfare facilities may register starting Monday through their facility's coordinator. General participants may sign up on a first-come, first-served basis starting at 10 a.m. on Nov. 23. "Access to advanced science and technology should not vary depending on where a child lives or their circumstances," National Science Museum Director Lee Eun-young said. "We will keep connecting online content with hands-on experiences so that children can continue to encounter science in their everyday lives."
Nov. 6, 2026
-
UNIST to drive defense AI innovation with Navy, Marine Corps
Busan AX hub opens, covering technology development and talent training Ulsan National Institute of Science and Technology (UNIST) said Friday it had opened a Navy and Marine Corps AI transformation hub in Busan to develop military-specific AI technologies and train specialized personnel. The new Navy and Marine Corps AX (AI transformation) hub is one of five national defense AX centers being established across the country to advance defense AI technology and build a pipeline of expert personnel. The five defense AX hubs are each tailored to the characteristics of their respective branches — the Navy and Marine Corps hub in Busan, the Joint Chiefs of Staff hub in Yongsan, Army hubs in Pangyo and Daejeon, and an Air Force hub in Yangjae. UNIST will lead the maritime domain, supporting AI technology development and field deployment for the Navy and Marine Corps. The Busan hub will operate as a collaborative platform combining the university's AI research and education capabilities with the military's data and operational experience and industry expertise. A particular focus will be placed on developing foundational AI technologies tailored to the Navy and Marine Corps and building a testbed to apply them in actual defense settings. Training personnel who can deploy AI in real military missions is another core objective. The hub plans to leverage the university's education and research strengths to secure defense AI specialists and to establish a cooperative framework that feeds the military's technology needs back into research and development. UNIST will also align its efforts with Busan's push to build a maritime AI belt. The plan envisions a collaborative ecosystem in which the region's AI research capabilities and industrial base are channeled into defense and maritime technology development, with research outcomes feeding directly into the field. The opening ceremony, held under the theme "A maritime powerhouse opened by AI, the AI transformation beginning in the maritime capital," featured a vision declaration, a plaque unveiling, and exhibitions and demonstrations of key research projects. UNIST President Park Jong-rae said the Navy and Marine Corps AX hub would serve as "a hub connecting the university's research achievements to solutions for the defense field," adding that UNIST would "expand its research and education capabilities into the defense and maritime sectors on the foundation of military-industry-academia cooperation."
Nov. 6, 2026
-
POSTECH professor wins minister's commendation for turning lab tech into medical devices
Kim Cheol-hong recognized for developing core technology behind photoacoustic-ultrasound imaging systems A South Korean researcher who developed next-generation medical imaging technology combining light and ultrasound — and carried it from the laboratory to commercialization — has received a government commendation. Pohang University of Science and Technology (POSTECH) announced Friday that Kim Cheol-hong, a professor in the departments of electrical engineering, biomedical engineering and mechanical engineering and in the convergence graduate school, received a commendation from the deputy prime minister and Minister of Science and ICT at the International Research Industry Convention 2026, held Wednesday. The commendation was established to identify and recognize individuals who have contributed to the promotion and development of the research industry. Kim was honored for developing photoacoustic-ultrasound fusion imaging equipment that combines the strengths of light and ultrasound, and for translating that work into the production of research instruments and the launch of a commercial venture. Photoacoustic imaging works by directing light at biological tissue; the tissue absorbs the light energy and generates faint ultrasound signals that are captured and rendered as images. When combined with conventional ultrasound imaging, the technology can reveal not only the structural information of tissue but also the optical properties of biological structures such as blood vessels. Kim has focused on securing core technology for an integrated imaging system that brings together optics, ultrasound, electronics and image processing, and on realizing that system as practical medical imaging research equipment. To apply the technology developed in his laboratory to industry, Kim founded Opticho, a company specializing in medical imaging. Building on photoacoustic imaging technology, the company has pursued the development and commercialization of preclinical and clinical research instruments, contributing to the technological competitiveness of South Korea's advanced medical imaging equipment sector. The award is significant because foundational university research has gone beyond equipment development to spawn a startup and bring products to market. "This award carries great meaning in that it recognizes a full-cycle achievement — one that did not stop at developing research equipment but extended to founding a company and commercializing products," Kim said. "Going forward, I will work to advance photoacoustic-ultrasound fusion medical imaging technology and expand the commercialization of research equipment, contributing to technological innovation in the advanced bio-health sector."
Nov. 6, 2026
-
South Korea moves to domestically source key parts for 30-ton armored vehicles
Korea Institute of Materials Science supports localization of drivetrain components for next-generation export armored vehicles South Korea is accelerating efforts to domestically source the critical drivetrain components that power 30-ton armored vehicles. The Korea Institute of Materials Science (KIMS) announced Friday that it has partnered with TIC, a Changwon-based defense drivetrain parts specialist in South Gyeongsang Province, to provide materials technology support for localizing key drivetrain components in next-generation export armored vehicles. The initiative targets domestication of drivetrain modules used in 30-ton, 700-horsepower next-generation armored vehicles — developing the components that transmit power and ensure driving stability using domestic technology, with the aim of reducing dependence on foreign suppliers. The push comes as South Korea's defense exports have surged in recent years, making a stable supply chain for critical components an increasingly urgent priority. According to the Defense Acquisition Program Administration (DAPA), domestic defense export orders reached $15.4 billion last year, up more than 60 percent from the previous year. Relying on foreign suppliers for critical components in major weapons systems such as armored vehicles, however, risks export approval delays and parts supply disruptions that can hamper delivery. Localizing not just finished weapons systems but also the underlying materials and components is essential to maintaining export competitiveness. The project brings together a large defense contractor overseeing armored vehicle system development, TIC handling the core drivetrain components, and specialized partner firms responsible for controllers and frames. The biggest obstacle to localization is securing high-performance materials capable of withstanding heavy loads, intense friction and extreme temperatures. Drivetrain parts such as gears and axles must endure the weight of a vehicle body weighing tens of tons as well as powerful engine output. A particular challenge has been that the material composition and heat-treatment conditions of existing foreign components are not disclosed, making domestic substitution difficult. KIMS has mobilized a team led by Won Soon-ho, a principal researcher at its Industrial Technology Solutions Center, to provide tailored materials technology support. Seven core components are targeted for support, including the auxiliary transmission, differential gear box and wheel hub. KIMS plans to conduct precise analysis of the material composition, microstructure and heat-treatment condition of foreign parts, then select substitute materials that meet domestic industrial standards. The review will cover not only material composition but also hardenability, cleanliness, post-heat-treatment mechanical properties and domestic procurement feasibility. Researchers also plan to analyze the causes of component failures that may arise during prototype production to improve reliability. KIMS aims to complete materials analysis and property evaluation of the key components and propose domestically sourced substitute materials by the end of the year. TIC is expected to benefit by securing the technical data needed for prototype development and component approval by vehicle manufacturers, while cutting development time and testing and analysis costs. The support is being carried out as a "cluster-type technology assistance" model — going beyond solving one company's technical difficulties to address the shared technology needs of multiple firms within the same defense supply chain. KIMS plans to extend the results beyond the defense sector to high-load, high-output vehicle applications such as independent suspension axles for special-purpose vehicles. "The materials technology and research infrastructure of government-funded research institutes can translate into industrial competitiveness only when they solve the real technical challenges companies face on the ground," KIMS Director General Choi Cheol-jin said. "We will strengthen the export competitiveness of K-defense by providing technology support that spans the entire supply chain of the region's key industries, including defense and mobility."
Nov. 6, 2026
-
Kakao releases AI decision model 43 times faster than GPT-6 Luna
Model delivers results as probabilities, choices and scores Speed clocks in 43 times faster than GPT-6 Luna Kakao has unveiled a new AI model designed to support decision-making by AI agents, with faster processing speeds aimed at improving operational efficiency. The company said Friday it released Kanana-2-v-decide, a multimodal AI model developed in-house, as open source on Hugging Face, the global AI platform. The model was built to handle the rapid judgments AI agents must make while carrying out tasks. Unlike conventional large language models, which respond to user queries in full sentences, it delivers results in a structured format. AI agents must make a continuous series of decisions — which tools to use, which information to reference — as they fulfill user requests. The model supports that process by supplying the judgment needed at each step so the agent can move quickly to the next action. Kakao built the model on Kanana-2-v-3b, a multimodal language model it had previously developed. Because it can analyze text and images together, it can be applied to a range of tasks including customer inquiry classification, tool selection by AI agents and product image assessment. The model analyzes input and returns results in three formats: probability, which indicates the likelihood that a given condition is true; choice, which identifies the most suitable option from a set of candidates; and score, a numerical value reflecting the probability at each evaluation stage. Kakao also improved processing speed by applying a technique that represents input queries and answer choices as vectors — numerical groupings — and compares their similarity. On a single Nvidia H200 GPU, the model recorded an average response time of under 0.04 seconds, about 43 times faster than GPT-6 Luna. The model also demonstrated competitive performance in benchmark evaluations. It ranked highest among all models tested on both Kanana Decision Bench — Kakao's proprietary evaluation dataset comprising 15,909 Korean-language questions — and Decision Index, a publicly available benchmark. Models compared in the evaluation included Microsoft's Decision-1, Typesafe AI's Jev, Perplexity's pplx-decider-v1.1-27B and Upstage's Solar Decide. Kakao applied its Kanana Open License to the model, permitting both research and commercial use. The company said the intent is to help researchers and businesses at home and abroad develop AI agent services. Kakao plans to deploy the model first in internal projects, including web search and automated product category classification. It aims to refine performance in live service environments and, over the longer term, develop technology that integrates search and decision-making into a single model. "This release is meaningful in that it broadens the foundation for applying cutting-edge AI technology to real services," said Choe Dong-jin, performance leader for applied AI models at Kakao. "We will continue to advance and open our technology so users can experience AI in their everyday lives." Meanwhile, on Oct. 29 Kakao released two embedding models — Kanana-2-embedding — as open source. The models, which underpin personalized AI agent technology, analyze the semantic similarity between user queries and information to surface relevant results, with a focus on improving Korean-language search performance and processing efficiency.
Nov. 6, 2026
-
OpenAI unveils GPT-6 Cyber, an AI model built to find software security flaws
Vulnerability-exploitation success rate hits 48.5%, up 16 percentage points from previous model Model used in Firefox, Linux security research Rolling out first to select Daybreak Red customers OpenAI has unveiled GPT-6 Cyber, an AI model designed to detect software security vulnerabilities and verify whether they can be exploited in practice. The model identified 81 previously unknown high-severity security flaws in the Firefox web browser. OpenAI announced Friday the release of GPT-6 Cyber, a cybersecurity-focused AI model with enhanced capabilities for discovering and verifying software vulnerabilities and supporting the development of security patches. A key feature is specialized training that keeps the model within task scopes and constraints approved by security researchers. GPT-6 Cyber will initially be available via API and Codex to select approved Daybreak Red customers, with plans to expand access to all Daybreak Red customers in the future. OpenAI highlighted the model's ability to identify vulnerabilities in large codebases and verify their real-world exploitability as key strengths, enabling it to generate patch candidates and analysis materials needed for future fixes. The company said the model is designed to help security teams find and remediate vulnerabilities before attackers can. OpenAI researchers used GPT-6 Cyber to uncover 81 previously unknown high-severity security vulnerabilities in Firefox, the web browser developed by Mozilla. The team generated verification materials for each flaw and shared them with Mozilla, which then released fixes based on the findings. In Linux kernel research, the model reproduced 1,724 vulnerabilities in total and developed working exploits for 948 of them. Following Linux community guidelines, OpenAI reported 304 critical findings to maintainers along with patch candidates, and work on fixes for additional vulnerabilities is ongoing. GPT-6 Cyber also outperformed its predecessor on major cybersecurity benchmarks. On ExploitGym, which evaluates whether vulnerabilities can be exploited in practice, the model achieved a success rate of 48.5% — about 16 percentage points higher than the 32.1% recorded by GPT-5.6 Cyber. In an initial evaluation on RemoteExploitBench, which tests complex security tasks on servers simulating real-world environments, the model completed 95% of the assigned challenges. OpenAI also strengthened the model's ability to stay within authorized task scopes. It received specialized training to adhere to approved targets and boundaries and to report results accurately, reducing the risk of a cybersecurity AI agent accessing unauthorized systems. GPT-6 Cyber also improved over its predecessor on ExploitGym Honeypot, a benchmark that assesses whether the model attacks services outside its authorized scope. In that evaluation, GPT-6 Cyber without deployment safety prompts refrained from attacking out-of-scope services in 99.93% of cases, compared with 94.49% for GPT-5.6 Cyber. In addition, OpenAI reinforced safeguards including monitoring to verify that the model's actions remain within the bounds of user-approved scope. The model is designed to block requests that could harm critical infrastructure or essential services while supporting approved security research and vulnerability verification.
Nov. 6, 2026
-
Seoul St. Mary's, KDCA share Candida auris infection control expertise with Japan
Hospital shares Candida auris management know-how with Japan KDCA, JIHS officials tour laboratory and ICU at Seoul St. Mary's Three-way framework to be built for cross-border disease response Seoul St. Mary's Hospital and the Korea Disease Control and Prevention Agency have shared their Candida auris infection control practices with Japan, as the two sides work to build a three-way cooperative framework to contain the spread of the fungal pathogen, which was recently designated a Class 4 notifiable infectious disease in South Korea. The hospital announced Friday that it had shared South Korea's sentinel surveillance system and best practices in hospital infection control and treatment with Japan's National Institute of Health and Medicine (JIHS), in collaboration with the KDCA, as a proactive measure against the spread of Candida auris. Candida auris is a fungus that poses significant challenges for treatment and infection control. It can cause invasive infections particularly in immunocompromised patients and those in intensive care, and has been associated with high mortality rates. Since its first reported case in South Korea in 2009, sporadic infections have continued, and the growing number of antifungal-resistant strains, invasive infections in healthcare settings and cluster outbreaks has heightened the need for closer surveillance and management. In response, the KDCA designated Candida auris a Class 4 notifiable infectious disease — specifically a healthcare-associated infection — effective March 29. On Wednesday, Seoul St. Mary's Hospital hosted a Korea-Japan infection control cooperation meeting and facility tour with experts from JIHS and KDCA officials. The visit was arranged in line with Japan's preparations to establish a national-level surveillance system, and aimed to strengthen cooperation with the KDCA while sharing the advanced infection control and diagnostic capabilities of a South Korean tertiary hospital. The visiting delegation included five officials from South Korean health authorities, among them Jeong Tong-ryeong, director general of the KDCA's Medical Safety and Prevention Bureau, as well as four specialists in infectious disease, diagnostics and epidemiology, including Ishikane Masahiro, an infectious disease specialist affiliated with JIHS. Seoul St. Mary's Hospital shared information on four areas: its Candida auris diagnostic and surveillance system centered on the clinical microbiology laboratory; patient management protocols and outbreak response cases; antifungal treatment options by individual patient case; and the status of antifungal susceptibility testing and genotyping of Candida auris strains conducted through a KDCA-commissioned private-sector project. The delegation then observed the hospital's integrated "diagnosis-isolation-treatment" control system, which covers Candida auris identification and carrier screening through to environmental management on wards and patient isolation. Health authorities and hospital representatives from both countries agreed to further strengthen their cooperative response to infectious diseases, using the visit as a springboard for deeper collaboration. "The most fundamental and essential element of patient safety at a tertiary hospital is rigorous infection prevention and control," said Kwak Seung-ki, deputy medical director of Seoul St. Mary's Hospital. "I hope this meeting serves as a meaningful opportunity to share the infection control expertise and clinical experience our hospital has accumulated with experts from Japan's health authorities." Ishikane Masahiro, the JIHS infectious disease specialist, said Japan is at a critical stage of building a national surveillance system for Candida auris and preparing its public health crisis response. "It is enormously meaningful to be able to learn firsthand about Seoul St. Mary's advanced infection prevention and control framework, its surveillance network operations, and its clinical experience with treatment options for multidrug-resistant patients," Ishikane said. Meanwhile, Seoul St. Mary's Hospital was designated a public safety hospital by the Ministry of Health and Welfare during the 2015 MERS outbreak, during which it operated a dedicated public safety clinic. During the COVID-19 pandemic, it was simultaneously designated both a Seoul Metropolitan Government COVID-19 severe emergency care center and a public safety hospital.
Nov. 6, 2026
-
Games, security software become conduits for malware as hacking threats spread
PC launcher for mobile game Goddess of Victory: Nikke hacked, used to distribute malware Security solution cPGuard's official update system also compromised Users may be exposed to cyberattacks without knowing it, raising urgent calls for end-to-end security Amid an expanding wave of hacking threats worldwide, even official programs for games and security software have been exploited as channels for distributing malware, setting off alarm bells across the industry. Attackers are embedding malicious code into normal distribution processes — such as launching a game or updating a security program. As hacking traps appear in the routines of daily digital life, concerns are growing that users may be exposed to cyberattacks without ever realizing it. According to the gaming and security industries, the PC launcher for Goddess of Victory: Nikke, a mobile game serviced domestically and internationally by China's Tencent, was hacked and used to distribute malware. Nikke was developed by South Korean game studio Shift UP and is distributed by Level Infinite, a Tencent subsidiary. The game has accumulated about 45 million downloads worldwide. The malware distribution stemmed from the tampering of a JavaScript file in the content delivery network used by the Nikke PC launcher on Sept. 27 by a third party. Hackers embedded malicious code into a file required to run the game, meaning users were exposed to malware simply by launching it as usual. The game's operators believe distribution of the malicious program through the compromised script began around 3:52 a.m. on Oct. 30. The malware was found to be capable of giving attackers remote control over a user's PC, including the ability to access files and execute commands. Level Infinite issued a belated apology, saying it had "completed the necessary security measures and fixes for the affected PC launcher" and had "prepared in-game compensation as an apology for causing users significant concern and worry." The game's operators have reported the breach to the Korea Internet & Security Agency, which is investigating the full extent and scale of the damage. Security software — the very tools designed to prevent hacking — has also been exploited as a vehicle for spreading malware. On Monday, the official update system for global security solution cPGuard was hacked. Attackers stole the access key used to distribute updates and replaced legitimate update files with malicious scripts. The incidents have heightened a sense of crisis that even official programs are no longer safe havens, pushing the industry into emergency response mode. The gaming sector moved quickly to limit further damage. Nexon issued a security notice instructing users to delete any folders related to the Nikke launcher from their PCs and contact its security team. Netmarble isolated PCs confirmed to have Nikke installed. NCsoft, Krafton and Smilegate are also checking their own security systems and deleting related files. The security industry is worried that the exploitation of security programs themselves as attack vectors could undermine trust in existing security frameworks. Particularly concerning is the difficulty of distinguishing legitimate programs from those carrying malicious code, which raises the risk of damage spreading widely. Calls are growing for urgent security reviews and stronger oversight across the entire lifecycle of software — from development and distribution to updates.
Nov. 6, 2026
-
KT Corp bolsters network ahead of Busan fireworks festival
19 temporary base stations to operate around Gwangalli Beach 5G technology to support 15 safety-monitoring CCTV cameras KT Corp is reinforcing its telecommunications network ahead of the Busan Fireworks Festival to handle the large crowds expected at the event. The company plans to install temporary base stations at the venue and apply traffic-distribution technology to provide stable service to attendees. KT Corp announced Friday it would activate a special communications plan for the 21st Busan Fireworks Festival, set for Saturday. The festival is expected to draw large crowds to Gwangalli Beach and surrounding areas in Busan. Mobile data usage is forecast to surge as more attendees share photos and videos taken at the event. In response, KT Corp conducted advance checks on network quality at Gwangalli Beach and along major pedestrian routes, and reinforced its wireless network accordingly. The company will operate 19 fixed temporary base stations and mobile base stations around Gwangalli Beach and nearby viewing zones. Additional 5G and LTE equipment has been deployed at beach access points and areas expected to see the heaviest crowds, with antenna angles and output levels optimized for on-site conditions. On the day of the event, KT Corp will run a real-time response system centered on its network operations center and an on-site command post. Staff will closely monitor traffic fluctuations and base station status before and after the event, with specialists on the ground ready to respond quickly to any issues. The company will also deploy W-SDN (Wireless Software Defined Network), a wireless network automation technology that redistributes load and optimizes the wireless network when traffic concentrates on specific base stations, minimizing any drop in service quality. KT Corp is also preparing for unexpected power failures. The company said it has standby power generation equipment ready and will manage both network quality and facility safety through the end of the event. The company is also stepping up communications support for crowd safety monitoring. Using 5G standalone (SA) network slicing technology, KT Corp will support video transmission from 15 wireless closed-circuit television cameras. The cameras will be installed at points of high crowd density, including Gwangalli Beach and subway station entrances and exits. KT Corp plans to allocate dedicated communications resources to ensure stable video transmission to the on-site command post. Network slicing divides a single 5G network into separate channels according to service requirements, each delivering the necessary quality of service. The technology allows KT Corp to secure dedicated communications resources for safety-monitoring video feeds even when general data usage spikes. For this festival, KT Corp plans to expand its coverage beyond outdoor venues such as Gwangalli Beach to include underground transit corridors connecting Gwangalli Station and Suyeong Station. "We reinforced the wireless network at the venue and along key pedestrian routes in advance so that festivalgoers can share their precious moments without any inconvenience," said Jeong Seon-il, senior vice president and head of network operations innovation at KT Corp's network division. "The operations center and on-site command post will work in close coordination to ensure stable communications service through the end of the event." Meanwhile, KT Corp took similar steps ahead of World Cup street cheering events in June, deploying special network management at major locations including Gwanghwamun Square, Seoul City Hall Plaza and Yeongdong-daero in Gangnam. On that occasion, the company used W-SDN technology to automatically manage base station overloads and prepared for large crowds through mobile base station deployment and round-the-clock monitoring.
Nov. 6, 2026
-
ETRI robot wins world competition by navigating unknown buildings without a map
- Robot independently determines exploration routes and return timing - Technology expected to find use in warehouses, disaster sites A robot that can find its own way through an unfamiliar building — and decide on its own when to head back to base — has taken first place at a major international competition. The autonomous exploration robot technology was developed by South Korean researchers. The Electronics and Telecommunications Research Institute (ETRI) announced Friday that it won the single-robot category at an international indoor autonomous exploration competition held alongside IROS 2026, a leading global robotics conference, in Pittsburgh in September. The competition evaluated how efficiently a robot could explore an unfamiliar indoor space within a set time limit and relay the information it gathered back to a base station. The final assessment used 10 indoor environment maps that had not been disclosed to participants, with four different starting positions applied in each map. The key performance metric was "base-station coverage" — not simply how much ground the robot covered, but the proportion of spatial information it successfully transmitted back to the base station. The ETRI team recorded a base-station coverage rate of 40.26 percent in the final evaluation, topping the single-robot category. Among the 10 finalists, the team posted the highest scores not only in the overall standings but also in both the small-map and large-map groups. The winning edge came from an autonomous exploration algorithm that lets the robot decide for itself where to explore and when to return to base. The ETRI team designed the algorithm to calculate travel distance and the amount of new information a given area could yield simultaneously, directing the robot to prioritize the most exploration-efficient locations. The system also factors in remaining exploration time and the distance needed to return to base, allowing the robot to cover as much ground as possible while still making it back within range. Another notable feature is that the technology requires no separate training data or pretrained AI model, meaning it can operate in unfamiliar environments without any additional learning. In internal comparison tests using sample competition environments, the baseline algorithm averaged a base-station coverage rate of 33.4 percent. Applying ETRI's technology raised that figure to 61.8 percent — an improvement of 28.4 percentage points. The technology is expected to be deployed in settings where obtaining maps in advance is difficult, including warehouses, large industrial facilities, below-ground spaces and disaster sites. "This victory is international recognition of the excellence of the autonomous exploration and path-planning technology our team has developed," said Byeon Woo-jin, head of ETRI's Daegyeong Regional Research Division. "Technology that allows robots to make independent judgments and explore on their own, even in environments without prior maps, will find applications across a wide range of fields — from logistics and industrial facilities to disaster response."
Nov. 6, 2026
-
Nexon to launch 'Temppalgye: Overgeared' on Nov. 26
Launching in South Korea, Taiwan, Hong Kong and Macao Game features equipment-driven growth and cooperative combat content Nexon is set to release a new game based on the popular web novel and webtoon series Temppalgye. The title brings to life the original work's signature equipment-driven growth system while also offering cooperative and competitive play between users. Nexon announced Friday that it will officially launch Temppalgye: Overgeared, a massively multiplayer online role-playing game developed by Gray Games, in South Korea, Taiwan, Hong Kong and Macao on Nov. 26. Temppalgye: Overgeared is a PC and mobile MMORPG developed from the intellectual property of the hit web novel and webtoon Temppalgye. The game centers on the core appeal of the original — growing stronger through equipment. On Thursday, Nexon held an online showcase to announce the official launch date and reveal the game's key content. Gray Games producer Choi Byeong-cheon and director Yun Sun-jun attended the event. At the heart of the game is a system in which players strengthen their characters by upgrading, crafting and modifying equipment. Through the Overgeared mechanic, players enhance slot options on individual gear pieces to boost their character's stats. Players can also engage in combat using Overclock, a feature whose effects vary depending on a character's equipment level, and God Hand, drawn from the original work. The game also introduces weapon modification, which alters a weapon's appearance and performance, and signature equipment, which bears the name of its creator. Large-scale cooperative content and cross-server interaction are also part of the experience. In the Inma Daejeon mode, all users on a server join forces to repel waves of monsters. The game will also roll out a Dimension system — allowing users from different servers to form parties or compete in player-versus-player battles — and Dimension League, which tallies individual contributions to determine server-wide rankings. "Temppalgye: Overgeared is an MMORPG where players do more than just grow stronger through equipment — they cooperate and compete with others to write their own stories," Gray Games producer Choi Byeong-cheon said. "We look forward to users beginning new stories of their own in the world of Temppalgye: Overgeared on Nov. 26." Meanwhile, pre-registration for Temppalgye: Overgeared has been open since Oct. 1 through the official website, Kakao, Google Play and the App Store. Users who pre-register will receive various rewards, including a Petal Rabbit companion pet and 10 confirmed alchemy catalyst boxes.
Nov. 6, 2026
-
KAIST researcher wins EU grant to develop brain-implant chip that could move wheelchairs by thought
- Professor Jang Kyung-in selected for European Research Council's Synergy Grant 2026 - Team to build 'brain map' by analyzing primate neural signals and behavior Could a tiny chip implanted in the brain one day let people control a smartphone or move a wheelchair using thought alone? KAIST announced Friday that NeuroEthoPrim, an international joint research project co-led by mechanical engineering Professor Jang Kyung-in, has been selected for the European Research Council's Synergy Grant 2026. It marks the first time a researcher affiliated with a South Korean university has been named a principal investigator on an ERC Synergy Grant. The project will receive about 9.68 million euros ($10.54 million) over six years, of which approximately 5 billion won has been allocated to KAIST. The ERC Synergy Grant is a large-scale basic research program through which the EU supports two to four world-class researchers working together to tackle scientific challenges that no individual researcher could solve alone. The project brings together Luca Bonini, a professor at the University of Parma in Italy; Silvestro Micera, affiliated with the Sant'Anna School of Advanced Studies in Italy and the Swiss Federal Institute of Technology Lausanne (EPFL); and KAIST's Jang. At the heart of the research is understanding how the brain functions while a subject moves freely. The team will build an experimental space called the NeuroEthoRoom, where monkeys can walk and forage freely. There, researchers plan to simultaneously collect brain activity data, whole-body movement data and tactile information to analyze how the brain translates sensory input into real-world action. The vast dataset gathered will be used to develop an AI foundation model that learns the relationship between neural signals and movement. Jang will oversee the development of a wireless neural interface — a device that reads brain signals or delivers electrical stimulation to specific regions of the brain — as part of the project. The team plans to develop technology that bridges the physical mismatch between soft brain tissue and rigid semiconductors, enabling stable recording of brain signals and precise stimulation of target areas even during free movement. The team's ultimate vision is a "Brain-to-X" platform — a single brain interface that connects to smartphones, prosthetic limbs, wheelchairs, hospital beds and residential spaces. The long-term goal is to allow patients to move a wheelchair or operate a smartphone by thought alone, and to receive tactile feedback while grasping objects with a robotic prosthetic hand. However, applying the technology to actual patients will require primate validation, human-subject research, safety and efficacy assessments, and regulatory approval. "Understanding how the brain links sensation and movement during free motion is the starting point of this research," Jang said. "We will do our utmost to ensure that the findings from this basic research help people living with neurological damage return to their daily lives."
Nov. 6, 2026
- 1Malaysia's BE International brings 1,500-strong incentive tour to Korea, bypassing Turkey
- 2AI singer tops Melon chart for first time with 'Did You Become the Adult You Dreamed Of?'
- 3Jensen Huang's daughter took the photo of him embracing his alleged mistress
- 4Man beat live-in girlfriend 514 times over 3 hours, killing her; forensic expert says footage was too disturbing to watch
- 5Madison Huang acted as de facto chief of staff at Nvidia, shouted at vice president: report
- 6'A concert is a ritual for the soul': Currentzis makes long-awaited Korea debut after six years
-
WORLD
US warns of punishment for anyone enabling Iran's Strait of Hormuz toll scheme
-
INDUSTRY
Toyota union puts productivity first; Hyundai Motor union demands bigger share of profits
-
FINANCE
National Growth Fund to invest $531M in FuriosaAI, which rebuffed Meta's $800M takeover bid
-
INDUSTRY
Korea Shipowners' Association holds amateur baduk tournament for Maritime Day
