United News of Bangladesh

DeepSeek’s chatbot experienced a major outage of more than seven hours overnight in China, prompting the AI firm to release multiple updates to fix the problem. Users first reported issues on Sunday evening, according to Downdetector. DeepSeek’s status page noted an initial disruption at 9:35 p.m., which was marked resolved two hours later. However, further performance problems emerged on Monday, taking until 10:33 a.m. to fully address. The exact cause of the outages remains unclear, and DeepSeek did not immediately respond to requests for comment. Extended downtime is rare for DeepSeek, a globally used AI app that has maintained nearly a 99% uptime record since the launch of its R1 model in January 2025. The company has been recognized as one of China’s leading AI innovators. Industry speculation suggests that DeepSeek, headquartered in Hangzhou, may be preparing a major update, following its high-profile debut on January 20 last year. The anticipated rollout has fueled competition, prompting rivals such as Alibaba Group, ByteDance, and Tencent Holdings to release new AI models over the Lunar New Year holiday. While anticipation for DeepSeek’s next move remains high, the company has not disclosed a timeline.  #From Bloomberg

As Iranian missiles hit Israel, some Android users received text messages promising real-time updates on nearby bomb shelters. But instead of helpful information, the links installed spyware, giving hackers access to cameras, location data, and personal information. The attack, linked to Iran, shows how cyber operations are now a key part of modern warfare. Experts say Tehran and its allied groups are using digital tactics to make up for military disadvantages, combining hacking, disinformation, and artificial intelligence. Gil Messing, chief of staff at cybersecurity firm Check Point Research, said the texts were timed to coincide with missile strikes, creating a “digital-physical” attack. “This was sent to people while they were running to shelters,” he said. “The exact timing is unprecedented.” Even if a ceasefire is reached, cyberattacks are expected to continue because they are cheap, fast, and focus on spying, theft, and intimidation rather than outright destruction. High-volume, low-impact attacks Most attacks so far have caused little direct damage but forced U.S. and Israeli companies to patch security weaknesses. DigiCert, a Utah-based cybersecurity firm, has tracked nearly 5,800 attacks by about 50 Iran-linked groups targeting networks in the U.S., Israel, and Gulf countries. Many attacks aim to intimidate rather than inflict major damage. Recently, a pro-Iranian group claimed responsibility for breaching an account of FBI Director Kash Patel, posting old personal documents. Such attacks are often designed to boost supporters’ morale and unsettle opponents. Hospitals and data centers under threat Iran is likely to target weak points in U.S. infrastructure, including hospitals, supply chains, and critical data centers. Recent strikes included Michigan-based medical company Stryker and another unnamed healthcare firm, hit with ransomware that aimed to disrupt rather than demand money. Cynthia Kaiser of Halcyon said, “There is a deliberate focus on the medical sector, and targeting is expected to increase.” AI’s role in cyber warfare Artificial intelligence is speeding up attacks and spreading false information, including deepfakes. One fake image of sunken U.S. warships received over 100 million views. Iranian authorities also control internet access to shape domestic perceptions of the war, sometimes labeling real footage as fake. In response, the U.S. created a Bureau of Emerging Threats last year to counter risks from new technologies. AI also helps defenders respond faster, according to Director of National Intelligence Tulsi Gabbard. While Russia and China remain the largest cyber threats, Iran has shown it can target American systems, including political campaigns, water plants, military networks, and online movements opposing Israel.

Yahoo is turning to artificial intelligence with its new answer engine, Scout, in a fresh attempt to regain its position in online search. The AI-powered tool provides direct answers along with links to supporting sources. In a response to an AP query, Scout said Yahoo’s decline showed how early success can fade without constant innovation. Yahoo CEO Jim Lanzone hopes to use AI to tap into the company’s global base of about 700 million users who still rely on its finance, sports, news and email services despite years of setbacks. Lanzone took charge after Apollo Global Management acquired Yahoo for $5 billion in 2021, far below its peak value of $125 billion during the early 2000 dot-com boom. Before that, Verizon had bought Yahoo’s core business in 2017 but failed to integrate it successfully with AOL. Years of missteps under multiple leaders weakened Yahoo’s standing, though it managed to survive, unlike some former tech giants, analysts say. Since taking over, Lanzone has focused on cutting underperforming units, selling assets like TechCrunch and shutting down AOL’s dial-up service. He says Yahoo is now profitable and generating billions in revenue. The company has also upgraded key products, including its fantasy sports platform and email service, which remains the second largest after Gmail. With Scout now rolled out to 250 million users in the US, Yahoo aims to offer simpler and more personalised search results. However, it faces tough competition from Google and AI platforms like ChatGPT, Claude and Perplexity. Yahoo is currently using AI technology licensed from Anthropic to run Scout. Lanzone said the tool is designed to deliver answers without mimicking human conversation. Founded in the 1990s as a web directory, Yahoo lost its edge after shifting focus away from search, allowing Google to dominate the space.

Melania Trump drew attention at a recent education and technology summit in Washington, but this time she shared the spotlight with a humanoid robot. On Wednesday, the former first lady attended the final day of a global summit held in the White House East Room, organized under her “Fostering the Future Together” initiative. The event brought together international representatives to explore how education, innovation, and technologies like artificial intelligence can help empower children. Melania Trump entered the venue walking alongside the robot, both moving slowly down a red carpet. Just before entering the East Room, she paused while the robot continued forward, circling a table of panelists before stopping at the center of the room. After briefly scanning the audience, the robot introduced itself as “Figure 03,” a humanoid created in the United States. It expressed gratitude for being invited and highlighted its role in supporting efforts to advance children’s education through technology. The robot also greeted attendees in multiple languages before exiting the room the same way it had entered. Melania Trump later thanked the robot, joking that it was her first American-made humanoid guest at the White House. The robot, developed by California-based Figure AI, was unveiled in October 2025 as a third-generation model designed to assist with everyday household chores such as cleaning, laundry, and dishwashing. Figure AI’s CEO, Brett Adcock, said he was proud to see the robot become the first of its kind to appear at the White House. The company is among several competitors—including Boston Dynamics, Tesla, and firms in China working to develop advanced human-like robots capable of performing practical tasks.

A court in New Mexico has ordered Meta Platforms to pay $375 million in damages after a jury found the company misled users about the safety of its platforms for children. The verdict followed a lawsuit brought by New Mexico Attorney General Raul Torrez, who described the ruling as “historic” and said it marked the first successful case by a US state against Meta over child safety concerns. The jury concluded that Meta — which owns Facebook, Instagram and WhatsApp — violated the state’s Unfair Practices Act by misleading the public about the risks faced by young users. Jurors found that the company’s platforms exposed children to sexually explicit content and contact with predators. The case was heard over seven weeks, during which jurors reviewed internal company documents and heard testimony from former employees indicating that Meta was aware of such risks. Among them was whistleblower Arturo Béjar, who told the court that his internal experiments showed underage users on Instagram were being served sexualised content. He also said his own daughter had received inappropriate sexual advances from a stranger on the platform. Prosecutors also presented internal research suggesting that at one stage, 16% of Instagram users reported encountering unwanted nudity or sexual activity within a single week. Meta, led by chief executive Mark Zuckerberg, rejected the findings and said it plans to appeal the decision. A company spokesperson said Meta continues to invest in safety measures and acknowledged the challenges of identifying harmful content and bad actors online, while maintaining confidence in its efforts to protect young users. The total penalty of $375 million was calculated after the jury determined there had been thousands of violations, each carrying a potential fine of up to $5,000. Meta argued that it has taken steps in recent years to improve user safety, including launching “Teen Accounts” on Instagram in 2024 to give younger users greater control, and introducing a feature last month to alert parents if their children search for self-harm-related content. The company is also facing a separate trial in Los Angeles, where a woman alleges she became addicted to platforms such as Instagram and YouTube — owned by Google — during her childhood due to their design. Thousands of similar lawsuits are currently pending across US courts. New Mexico filed the case in 2023, accusing Meta of directing young users towards sexually explicit material, child abuse content, and even solicitation and trafficking-related risks through its recommendation algorithms. “Meta executives knew their products harmed children, ignored warnings from their own staff, and misled the public,” Torrez said, adding that the jury’s decision reflects growing concern among families, educators and child safety advocates. #From BBC

Chinese robotics company Westlake Robotics has introduced its humanoid robot, Titan o1, driven by a self-developed large-scale model known as General Action Expert (GAE), capable of replicating human movements in real time without constraints of time or space. The company said the model allows multiple robots to carry out identical tasks simultaneously under the control of a single operator. During a demonstration in Hangzhou, an operator wearing a motion-capture suit performed actions such as waving, turning and kicking a ball, while Titan o1 mirrored each movement within milliseconds. The robot’s motions including arm swings, torso rotations, step length and leg lift during the kick remained closely synchronized with the operator’s rhythm. “These actions were instant responses to the operator’s spontaneous movements,” said Westlake Robotics founder Professor Wang Donglin, adding that Titan o1 can quickly adjust to different operators and variations in movement. He compared the GAE model to the human cerebellum, which coordinates movement, balance and precision, noting that it enables the robot to execute appropriate actions immediately upon receiving signals, even if the movements have not been performed before. Westlake Robotics also said the GAE model supports cross-embodiment deployment, meaning it can be applied to various robots with differing designs and sizes.

Ukraine’s front lines have evolved into a testing ground for rapid drone innovation, as soldiers develop low-cost technologies to counter waves of Iranian-designed drones used by Russia. Across eastern Ukraine, small military teams are working to intercept Shahed loitering munitions — relatively inexpensive but highly effective weapons that have reshaped modern warfare. Initially, Ukraine had limited means to counter such drones when they first appeared in 2022. Today, locally developed interceptor drones are increasingly used to destroy them mid-air. Soldiers from units such as the 127th Brigade continuously test and refine their equipment, often modifying reusable drones under battlefield conditions. These frontline efforts have contributed to the growth of Ukraine’s domestic drone industry, attracting attention from international partners. President Volodymyr Zelenskyy has said that countries in the Middle East have approached Ukraine for assistance in defending against similar drone threats. Military personnel say necessity drove the innovation. “We did it because we had nothing else,” said one pilot, describing the early days when traditional air defence systems struggled to track and intercept agile drones. Cost has been a key factor. A single Patriot missile can cost millions of dollars, while interceptor drones used by Ukrainian forces are far cheaper and can sometimes be reused, offering a flexible alternative. The shift toward drone-based defence began after conventional methods proved ineffective against fast-moving reconnaissance drones. Soldiers soon realised that the most effective counter to a drone was another drone. Experts say Ukraine’s experience highlights a broader escalation in warfare, where adaptable, low-cost technologies are increasingly shaping battlefield strategy.

A significant gap exists between parents and teenagers over the use of artificial intelligence (AI), with many parents unaware of how their children are engaging with the technology, according to recent studies. Surveys by Pew Research Center and Common Sense Media found that while 64% of teenagers reported using AI tools, only 51% of parents believed their children were doing so. Researchers said a lack of communication is a key concern, with four in 10 parents admitting they have never discussed AI use with their children. The studies showed that teenagers commonly use AI for schoolwork, research and entertainment, though some also turn to chatbots for personal advice and emotional support. About 12% of teens said they use AI for emotional guidance, while 16% use it for casual conversation, raising concerns among experts about overreliance on digital tools. Parents expressed unease over such uses, particularly when AI replaces human interaction, though many remain unaware of these behaviours. Experts warned that without proper guidance, teenagers may navigate AI independently, potentially leading to misuse or dependency. Despite concerns, many teenagers reported confidence in using AI tools and viewed them positively, with a significant proportion believing the technology would benefit them in the long term. Researchers urged parents to engage more actively with their children on AI use to better understand its role in their daily lives.   With inputs from BBC

Scientists in China have identified why people with chronic pain often feel more discomfort at night than during the day. The research, led by Zhang Zhi from the University of Science and Technology of China, was published Friday in the journal Science. Although it has long been known that pain follows a daily pattern—usually milder during active hours and stronger during rest—the exact reason was not clear. Scientists were aware that the brain’s internal clock, called the suprachiasmatic nucleus, controls sleep and hormones, but its role in pain was not fully understood. Using advanced techniques, researchers traced a specific nerve pathway in mice linking this brain clock to the spinal cord. They found that this pathway is influenced by the body’s natural daily rhythm. Since mice are active at night and rest during the day, the pattern is opposite to humans. During their resting period, the brain clock is more active, which increases pain signals. When the mice are active, the brain activity drops, reducing the intensity of pain. The findings help explain why pain sensitivity changes over the course of a day. Researchers say this discovery could help improve pain treatment by timing medications according to the body’s natural biological clock.

A senior vice president of and two associates have been charged in the United States with conspiring to smuggle billions of dollars’ worth of computer servers equipped with advanced chips to in violation of U.S. export control laws. Federal prosecutors said the defendants diverted large quantities of high-performance servers assembled in the U.S. to China between 2024 and 2025. Investigators allege they used fabricated documents, staged equipment to pass audits and relied on a pass-through company to conceal their activities and true customers. The accused include Yih-Shyan “Wally” Liaw, 71, a U.S. citizen and senior vice president and board member of Super Micro Computer; Ting-Wei “Willy” Sun, 44, a company contractor; and Ruei-Tsang “Steven” Chang, a Taiwan-based sales manager who remains at large. Liaw was arrested in California and released on bail, while Sun was held pending a bail hearing. According to court papers, Liaw and Chang directed a Southeast Asian firm to place about $2.5 billion in server orders from the California-based company, with at least $510 million later diverted to China. Super Micro said the alleged conduct violated company policies and that it is cooperating with investigators. Nvidia said it maintains strict compliance measures and does not support systems diverted in breach of export regulations.