分类: technology

  • China lands reusable rocket for first time, state media says

    China lands reusable rocket for first time, state media says

    China has marked a transformative milestone in its national space exploration program, achieving the first ever successful vertical landing and recovery of a reusable rocket booster, state-run media has confirmed.

    The mission, developed and executed by the China Aerospace Science and Technology Corporation, saw the Long March 10B rocket lift off from the coastal Wenchang Spacecraft Launch Site in China’s southern Hainan Province at 12:15 local time (04:15 GMT) on Friday. Approximately six minutes after the booster separated from the rocket’s upper stage, it completed a controlled vertical descent back to Earth and was successfully retrieved on a purpose-built floating recovery platform in the South China Sea.

    This landmark achievement opens a new chapter in global commercial spaceflight, positioning China as an emerging competitor to the long-held U.S. dominance in reusable rocket technology. For nearly a decade, American aerospace firms have led the sector: SpaceX, founded by entrepreneur Elon Musk, completed the first successful orbital reusable rocket landing in December 2015 with its Falcon 9 booster, while Blue Origin—owned by Amazon founder Jeff Bezos—followed with its own successful landing of the New Glenn rocket in November 2025. Today, the Falcon 9 conducts roughly 150 launches annually, with each booster capable of being reused more than 30 times, drastically cutting launch costs for commercial and government clients alike.

    Prior to this mission, China conducted its first experimental reusable rocket recovery test in February of this year, when the earlier Long March 10A prototype completed a controlled descent and splashed down near a recovery vessel, stopping short of a full vertical capture. The Long March 10B, which is capable of delivering a minimum payload of 16 metric tons to low-Earth orbit, is functionally comparable to SpaceX’s workhorse Falcon 9, though it employs a unique landing mechanism: unlike the Falcon 9, which lands autonomously on drone ships or ground pads, the Long March 10B uses landing hooks mounted to the booster to catch a reinforced net stretched across the floating recovery platform.

    For decades, the global space industry relied on expendable rocket designs, where all segments of the launch vehicle are discarded and destroyed during ascent, driving astronomical costs for every launch. By reusing rocket boosters—the single most expensive component of any launch vehicle—organizations can slash launch costs by up to 90% in some cases, making frequent satellite deployment, deep space exploration, and even commercial space tourism far more accessible.

    In reaction to the breakthrough, public shares of Chinese aerospace companies soared on domestic markets immediately after the successful landing was announced. Both China Spacesat and China Satellite Communications saw their share prices rise by 10%—the maximum daily gain allowed under Chinese financial market regulations.

  • Outcry as Meta lets users make AI images from public Instagram profile pics

    Outcry as Meta lets users make AI images from public Instagram profile pics

    Tech giant Meta has sparked widespread public and regulatory criticism over its newly launched artificial intelligence image generation tool, Muse Image, which carries the controversial capability to create AI-altered images using existing public profile pictures from users without their prior explicit consent.

    As a new addition to the fast-growing market of text-to-image generative AI tools, Muse Image converts short text prompts into custom-created images. It is currently rolled out to users across the United States, accessible via the Meta AI standalone app and web interface, as well as integrated into WhatsApp and Instagram Stories. Unlike many competing tools that rely exclusively on user-uploaded source material, Muse Image draws on the massive library of public user content hosted across Meta’s social platforms, enabling it to incorporate existing profile and user images into new AI-generated creations.

    While Meta has framed the tool as an accessible creative innovation, and confirmed that even users with public Instagram accounts can opt out of having their images repurposed for AI generation via a dedicated settings toggle, critics have decried the feature as a major threat to digital privacy and personal security. The opt-out process requires users to manually navigate to Instagram’s settings menu, select the “Sharing and Reuse” tab, and disable the content reuse toggle for posts and reels – a step many users are unaware of, according to campaigners.

    Donald Campbell, advocacy director at UK-based tech justice non-profit Foxglove, told the BBC that the unconsensual reuse of personal images is an “obvious recipe for disaster”. Campbell noted that over the past 12 months alone, social platforms have already recorded a long list of harms stemming from non-consensual AI-altered images, ranging from non-consensual deepfake pornography to identity fraud and reputational damage. He questioned why Meta CEO Mark Zuckerberg would choose to facilitate what he described as “creepy image manipulation” on a mass scale.

    Privacy International, another global digital rights advocacy group, also joined the criticism, arguing that the launch of Muse Image is the latest example of big AI companies treating users’ personal data and likenesses as uncompensated raw material for commercial exploitation. The criticism has been echoed by ordinary social media users, with one commenting on X that pulling real users into AI-generated photos without explicit permission is “a privacy landmine waiting to detonate”.

    The controversy comes as global regulators are already ramping up scrutiny of non-consensual AI image generation across the tech sector. UK communications regulator Ofcom is currently conducting an investigation into X (formerly Twitter) over the role of Elon Musk’s Grok AI in creating and distributing non-consensual AI-altered images of real individuals. Meta’s launch of Muse Image is almost certain to face intensified regulatory review amid growing concerns about the risks of unregulated generative AI.

    In its official description of the tool, Meta says Muse Image uses advanced AI reasoning to interpret complex user prompts, and can seamlessly blend multiple source images into high-quality finished creations that users can download and share across any platform. The tool also includes preset prompt suggestions to inspire creative ideas, and allows users to sketch custom edits directly onto generated images. While the core tool is free for casual everyday creative use, Meta offers expanded usage limits for heavy users via its paid subscription plans.

    A test of the tool demonstrated its capabilities: when asked to generate an image of the tester driving a car, the AI produced a convincing result – though it made a notable error, placing the steering wheel on the left side rather than the right, the standard for vehicles in the United Kingdom where the test was conducted.

    Looking ahead, Meta has confirmed that it plans to expand access to Muse Image in the coming months, rolling the feature out to Facebook and Messenger, and also making the tool available to advertisers for commercial content creation. A video generation variant of the tool is also reportedly in active development at the company.

  • Victims of 23andMe data breach to get $47m payout, judge rules

    Victims of 23andMe data breach to get $47m payout, judge rules

    One of the most high-profile consumer data breaches in recent biotech history has moved toward resolution, after a California bankruptcy court judge approved a $46.75 million compensation package for users affected by 23andMe’s 2023 genetic data hack.

    The court’s Tuesday ruling orders Chrome Holding, the entity that acquired 23andMe following the company’s 2024 bankruptcy restructuring, to disburse the £35 million (approximately $46.75 million) settlement within five business days to Kroll Restructuring, the court-appointed firm representing breach victims. Kroll will then handle distribution of funds to eligible users affected by the incident. This structured compensation process is standard for bankruptcy-related legal settlements, where third-party administrators manage claims on behalf of harmed parties.

    23andMe, the pioneering direct-to-consumer DNA testing firm that launched in 2006 and went public in 2021, built its business by creating detailed genetic profiles for millions of customers, holding deeply personal data ranging from disease predispositions to family ancestry connections. The 2023 breach began when hackers gained unauthorized access to roughly 14,000 direct user accounts. Because of 23andMe’s DNA matching service that connects users to biological relatives, hackers were able to expand their access to an estimated 6.9 million total user profiles, exposing the highly sensitive genetic information of millions of people who never had their own accounts directly compromised.

    The incident sparked widespread criticism of 23andMe’s lax data security practices. Regulators on both sides of the Atlantic brought penalties: the UK’s Information Commissioner’s Office (ICO) issued a £2.31 million fine, finding the company had failed to implement basic adequate safeguards for sensitive user genetic data. In May 2024, California Attorney General Rob Bonta filed a lawsuit against the firm, concluding an investigation that found 23andMe not only neglected basic security protocols to protect user information, but also intentionally misled consumers about the full scope and severity of the 2023 breach.

    The company’s financial struggles long predated the fallout from the hack. 23andMe never turned a profit over its nearly 20 years of operation, and at its peak reached a valuation of $6 billion. Last year, roughly 18 months after the breach was first disclosed, the company filed for Chapter 11 bankruptcy protection. Co-founder Anne Wojcicki acquired the company’s assets via a $305 million bankruptcy auction through her holding entity Chrome Holding, which operates under the trade name TTAM Research Institute. The company has continued normal operations following the restructuring, still selling DNA testing kits to consumers online.

    As of this ruling, it remains unclear exactly how many of the 6.9 million affected users will qualify to receive compensation from the settlement fund. Representatives for Chrome Holding, 23andMe, and the victims’ legal team have been contacted for additional comment on the settlement timeline and claims process.

  • Microsoft cuts 4,800 jobs and shrinks Xbox in ‘significant restructure’

    Microsoft cuts 4,800 jobs and shrinks Xbox in ‘significant restructure’

    Microsoft, one of the world’s largest technology corporations, has announced a round of company-wide layoffs affecting 4,800 positions, equal to roughly 2.1% of its total global workforce. The company’s Xbox gaming division is set to absorb the largest share of these cuts, in what executives describe as the most sweeping organizational reset in the brand’s decades-long history.

    In an internal memo distributed to all employees, Microsoft Executive Vice President Amy Coleman framed the layoffs as a necessary strategic adjustment to align the company with shifting market conditions amid a rapidly evolving technology landscape. Coleman emphasized that companies cannot halt industry-wide change, but only choose how to adapt alongside it. She confirmed that while the company will not fill eliminated roles with artificial intelligence, the growing adoption of AI is fundamentally reshaping how work is completed across all departments, requiring a leaner, more focused operational structure.

    For Xbox, newly appointed Chief Executive Asha Sharma outlined the full scope of changes in a staff note published to the social platform X. The restructuring will eliminate more than 1,600 roles immediately, with an additional 1,600 positions cut over the coming 12 months. As part of the broader reorganization, four Xbox-owned game development studios — Compulsion Games, Double Fine Productions, Ninja Theory and Undead Labs — will be spun out from Xbox to operate as independent entities, retaining full ownership of their existing intellectual property.

    Sharma pushed back against assumptions that the cuts signal a retreat from the gaming market, telling staff “these changes are about a bigger future for Xbox, not a smaller one.” She added that many legacy companies have mistakenly assumed long-term market dominance is guaranteed, and Xbox would avoid that complacency by proactively resetting its strategy to meet changing consumer expectations.

    As part of the reshuffle, two of Xbox’s highest-performing studios — Minecraft developer Mojang and Candy Crush creator King — will now report directly to Sharma, a shift that analysts say underscores the brand’s new focus on high-value, mass-audience intellectual property. The spin-out of the four smaller studios comes after Microsoft spent years acquiring independent studios to bulk up content for its Game Pass subscription service. Industry analysts note that the decision to spin the studios out rather than shut them entirely is a rare positive development amid a wave of widespread gaming industry layoffs.

    Representatives from the affected studios have publicly commented on the transition, with both expressing gratitude for their time under Xbox ownership and optimism for their independent futures. “We’re thankful to everyone at Xbox for seven great years together,” Double Fine, the studio behind the award-winning Psychonauts franchise, wrote on X. “This outcome preserves our history and culture, and returns ownership of our games to us.” Compulsion Games, developer of the upcoming action-adventure title South of Midnight, echoed that sentiment, noting its immediate priority is supporting its team through the transition and expressing confidence in its independent path forward.

    The latest layoffs come as the global gaming industry continues to grapple with widespread workforce reductions following years of post-pandemic contraction. In 2024 alone, Xbox cut more than 2,000 positions and closed four studios acquired ahead of its multi-billion-dollar purchase of Call of Duty developer Activision-Blizzard. Just over a year ago, Microsoft announced another 9,000 company-wide layoffs as it committed to doubling down on multi-billion-dollar investments in artificial intelligence development. Rising hardware production costs, driven in part by skyrocketing demand for AI data center infrastructure that has outpaced global supply chains, have also forced Microsoft and other consumer tech firms to raise prices on older consoles and gadgets.

    Industry analysts have described the restructuring as a defining moment for Xbox, as the brand navigates an increasingly fragmented gaming market that spans traditional consoles, personal computers, cloud streaming and subscription services. “The challenge is not just cutting costs; it is defining what Xbox stands for in a world where games are moving across console, PC, cloud and subscription platforms,” noted tech analyst Paolo Pescatore in comments to the BBC. Piers Harding-Rolls, head of games research at Ampere Analysis, added that the restructuring clarifies Xbox’s long-term vision: the brand will now prioritize its largest, most popular intellectual properties and global audiences, while smaller studio projects will have more room to grow outside the corporate structure.

    This reset marks one of the most significant strategic shifts for Xbox in its 24-year history, as the brand adapts to a new era of gaming shaped by AI innovation, changing consumer behavior, and ongoing industry-wide cost pressures.

  • Top EU court dismisses Google appeal of $4.5 billion antitrust fine

    Top EU court dismisses Google appeal of $4.5 billion antitrust fine

    BRUSSELS — In a landmark decision that solidifies European regulators’ leading role in reining in the power of big tech, judges at the European Union’s highest judicial body have rejected Google’s final appeal against a record 4.1 billion euro ($4.5 billion) antitrust penalty. The fine was originally issued over allegations that Google abused the dominant market position of its Android mobile operating system to stifle competition and narrow consumer options.

    The legal battle stretches back six years to 2018, when the European Commission first unveiled the penalty over Google’s business practices related to Android. Since that initial ruling, the case has wound its way through multiple layers of the EU’s judicial system, with Thursday’s judgment from the European Court of Justice bringing the long-running dispute to a definitive close.

    “The appeal filed by Google and its parent holding company Alphabet challenging the earlier ruling from the General Court is dismissed in full,” the Luxembourg-based court stated in its official ruling. This outcome formally upholds the penalty that was originally handed down for anti-competitive behavior tied to Google Search’s abuse of market power within the Android ecosystem.

    Android, an operating system Google positions as free and open-source, currently holds the title of the world’s most widely used mobile operating system, outpacing Apple’s iOS in global market share. Throughout the appeal process, Google maintained that its Android platform has driven down consumer prices for smartphones, expanded access to mobile technology, and created robust competition against its largest rival, Apple.

    This penalty against Google is just one of three major antitrust fines that the European Commission levied on the tech giant between 2017 and 2019, which together add up to more than $8 billion in total penalties. These early enforcement actions placed the 27-nation EU years ahead of other major global economies in the movement to curb anti-competitive practices by the world’s largest digital firms.

    In the years following those initial penalties, the European Commission has expanded its regulatory crackdown on big tech significantly, launching new antitrust investigations targeting other leading digital players including Amazon, Apple and Meta (formerly Facebook). It has also enacted sweeping new regulatory frameworks designed to specifically target the market power of the largest digital platforms, most notably the landmark Digital Markets Act.

    Agustín Reyna, director general of the European Consumer Organization, praised the court’s final judgment, noting that the ruling underscores the need for expanded proactive regulation like the Digital Markets Act to stop unfair corporate practices before they harm consumers and competition. “Today’s judgment sends a very clear message: dominant companies cannot use their power to shut out competition and limit consumer choice,” Reyna said. “Today is a big win for Europe.”

  • Google must pay €4.1bn fine for using Android to ‘block’ rivals

    Google must pay €4.1bn fine for using Android to ‘block’ rivals

    After years of legal wrangling between Google and European antitrust regulators, Europe’s highest judicial body has finalized its ruling: the U.S. tech giant must comply with a €4.1 billion (£3.5 billion) fine for anti-competitive behavior tied to its dominant Android mobile operating system. The decision dismisses Google’s latest appeal of the penalty, closing a years-long chapter in one of the most high-profile antitrust cases against Big Tech.

    The case traces its origins back to 2018, when the European Commission first leveled a €4.3 billion (£3.9 billion at the time) fine against Google, accusing the company of leveraging Android’s market power to freeze out competing products. The fine was adjusted down to €4.1 billion in 2022 as part of a procedural review, but Google continued to challenge the ruling before Europe’s top court. Wednesday’s dismissal of the appeal leaves the trimmed penalty intact, marking the largest penalty the European Commission has ever issued against Google to date.

    Regulators laid out three core illegal practices in the original 2018 ruling. First, Google forced manufacturers of Android-powered handsets and tablets to pre-install both Google Search and the Chrome web browser as a non-negotiable condition for gaining access to the Google Play app store, the primary distribution hub for Android applications. Second, the company offered financial incentives to major device manufacturers and mobile network operators that agreed to exclusively pre-install Google Search across their product lines. Third, Google blocked manufacturers from selling devices running modified, forked versions of Android by threatening to revoke their licenses to pre-install Google’s popular apps. Regulators did acknowledge that Google’s default Android setup does not prevent end users from manually downloading alternative browsers or search engines after purchasing a device.

    In a statement responding to the latest ruling, a Google spokesperson criticized the court’s decision, arguing it fails to acknowledge the company’s substantial investments in maintaining Android as an open, interoperable, and free operating system for developers, partners, and users. “In any event, we adapted our agreements to comply with the initial decision back in 2018 and we remain focused on continued innovation and openness for our users, partners and developers,” the spokesperson added. The company’s chief executive Sundar Pichai echoed this sentiment when the original fine was announced, writing in a 2018 blog post that the ruling threatened the core business model that has made Android a platform that expanded consumer choice rather than restricting it.

    This latest ruling is far from the only antitrust action Google and its parent company Alphabet have faced from European regulators in recent years. In 2024, the Commission ordered Google to pay a €2.4 billion (£2 billion) fine for abusing its market power in the shopping comparison service space, prioritizing its own offering over that of rivals. A year later, in 2025, regulators issued another €2.95 billion (£2.5 billion) penalty after finding Google broke competition rules by prioritizing its own ad tech products in search results, at the expense of competing ad providers.

    Notably, this €4.1 billion EU penalty is not the largest fine Google has ever been ordered to pay globally. In October 2024, a Russian court hit the company with an unprecedented fine of two undecillion roubles for restricting access to Russian state media channels on Google-owned YouTube, a sum that exceeds the total global GDP.

    Readers interested in tracking the latest developments in global technology policy and Big Tech regulation can sign up for the outlet’s Tech Decoded newsletter to receive top stories and trend analysis directly to their inboxes.

  • Alleged Scattered Spider hacker arrested in Finland

    Alleged Scattered Spider hacker arrested in Finland

    A years-long cross-border investigation into one of the world’s most disruptive modern hacking networks has led to the extradition of an alleged teen member to the United States, federal law enforcement officials confirmed this week.

    The US Department of Justice (DoJ) announced Tuesday that 19-year-old Peter Stokes, a dual US-Estonian national, has been charged with multiple felonies including conspiracy, computer intrusion, and wire fraud linked to his alleged role in the transnational hacking collective Scattered Spider. Stokes was first arrested by Finnish law enforcement in April following an Interpol Red Notice, and was transferred into US custody earlier this week. He made his first appearance at a Chicago federal court on Tuesday, where a judge ordered he remain in pretrial detention.

    Per DoJ allegations, Scattered Spider has carried out a years-long campaign of high-impact ransomware attacks that have generated more than $100 million in illegal ransom payments globally. The charges against Stokes build on recent progress in the case: just last month, two young men pleaded guilty to criminal charges connected to the 2024 cyberattack on Transport for London (TfL), a major breach UK investigators have directly attributed to Scattered Spider. That 2024 intrusion compromised the personal data of roughly 10 million TfL customers, causing £39 million in total damage. UK’s National Crime Agency (NCA) has also linked the group to separate 2023 cyberattacks on major British retailers Co-op and Marks & Spencer, which remain under active investigation.

    Court documents outline one specific attack linked to Stokes and his co-conspirators: in 2023, the group infiltrated the network of an unnamed luxury jewelry retailer, stole sensitive internal data, and demanded an $8 million ransom paid in cryptocurrency. According to the DoJ, the retailer successfully evicted the hackers from its network and refused to pay the extortion demand. Even so, the breach still caused at least $2 million in losses stemming from business disruption, forensic investigation, and threat mitigation work.

    The operation to arrest and extradite Stokes was a multinational collaboration, with joint work from FBI teams based in Copenhagen and Chicago, Finland’s National Bureau of Investigation, and Interpol. Security researchers and law enforcement have long noted that Scattered Spider is unusual among hacking groups for its membership: most members are believed to be young, native English speakers based in the United States and United Kingdom, a profile that has allowed the group to carry out socially engineered intrusions more effectively than many foreign-based criminal hacking networks.

  • India asks WhatsApp to pause username feature rollout over fraud concerns

    India asks WhatsApp to pause username feature rollout over fraud concerns

    India, WhatsApp’s largest single market with more than 850 million active users, has ordered the Meta-owned messaging platform to pause the rollout of its highly anticipated username-based chat feature, citing growing concerns that the update could fuel a surge in cybercrime and malicious online activity.

    The new function, which was set to roll out gradually to WhatsApp’s 3 billion global users over the coming months, allows users to connect with other accounts without sharing their personal phone numbers, a long-requested privacy upgrade for the platform. Users were already able to reserve their preferred unique usernames ahead of the full launch when Indian regulators stepped in.

    In an official notice dated Wednesday, India’s Ministry of Electronics and Information Technology (MeitY) demanded WhatsApp halt deployment immediately and explain why regulatory action should not be taken against the company under existing Indian cyber law for moving forward with a feature it argues poses unacceptable public risk. The ministry outlined that by allowing bad actors to contact potential victims without revealing their phone numbers, the update could materially increase rates of online fraud, phishing schemes, digital extortion scams, and impersonation attacks. Regulators also warned that the username system could enable bad actors to create lookalike accounts mimicking government officials, financial institutions, public agencies, and private individuals to carry out deception.

    MeitY’s notice, a copy of which has been obtained by the BBC, cites provisions of India’s Information Technology Act and national rules governing intermediary due diligence, identity theft, and impersonation offenses. The regulator has ordered WhatsApp not to proceed with launch until it completes a government consultation that addresses official concerns to the authorities’ satisfaction.

    In response to the order, Meta confirmed that the feature has not yet been activated for general users and that it has built a multi-layered framework of safeguards to prevent misuse. To combat impersonation, the platform has pre-reserved all high-profile usernames associated with public figures, government entities, celebrities, and already verified Meta accounts, ensuring these can only be claimed by their legitimate owners. It has also locked lookalike derivatives of prominent official and public figure usernames to block spoofing attempts.

    WhatsApp also emphasized that users will still require a verified phone number to create an account, even with the new username feature active. Additional protective measures include requiring any user to know an exact username to initiate contact, limiting the number of new users a single account can reach out to, blocking repeated automated attempts to guess usernames, and maintaining automated systems to detect and remove activity matching known patterns of impersonation and abuse. For first-time messages from unknown contacts, the platform will also display contextual information to help recipients assess risk, including whether the account is new, shares a common group with the recipient, is already saved in the user’s contacts, or is based in a foreign country.

    Cybercrime has become an increasingly urgent policy challenge for India, as hundreds of millions of first-time internet users adopt digital platforms and mobile payment services every year, many without formal training on digital safety practices. According to the most recent published federal data, India registered nearly 102,000 cybercrime cases in 2024, representing an 18% increase from the prior year, with online fraud accounting for nearly three-quarters of all reported cases.

    The Indian government’s order has drawn criticism from digital rights advocates, who argue the regulatory move lacks a clear legal foundation. The Internet Freedom Foundation, an Indian digital rights non-profit, stated that the notice overreaches by attempting to grant the government power to approve or reject individual software features, a authority not outlined in the laws MeitY cited. “The power to require prior permission for a feature is not in the [Information Technology] Act, not in the Rules, and cannot be created by a notice,” the organization said in a public statement.

    This latest regulatory action is part of a broader ongoing trend of Indian authorities tightening oversight of global technology platforms operating in the country. Earlier this year in February, New Delhi amended its digital rules to require social media platforms to remove unlawful content within just three hours of receiving a government takedown notice, cutting the previous compliance deadline of 36 hours dramatically. Just last month, Indian authorities temporarily banned the messaging platform Telegram ahead of a retest for the country’s national medical entrance examination, over concerns that its username-based interaction and hidden phone number features made it easier for exam cheaters to coordinate. The government’s position was upheld in court after Telegram unsuccessfully challenged the ban.

    For Meta, the regulatory clash in India carries particular weight: as WhatsApp’s largest market by user count, any changes to product rollout or design in the country will have a major impact on the platform’s global rollout plans and long-term growth strategy.

  • Bereaved South Koreans try AI-generated videos of deceased loved ones

    Bereaved South Koreans try AI-generated videos of deceased loved ones

    For 28-year-old South Korean office worker Lee Geon Hui, creating a meaningful gift for his single father meant turning to cutting-edge artificial intelligence to resolve a decades-old unspoken regret. Lee’s grandfather — the man his father had grieved for years after his unexpected death in a car crash before Lee was born — was brought back as a digital AI avatar to deliver a heartfelt personal message.

    Lee worked with Seoul-based AI startup Vaice in December 2024, crafting a script that reflected the words his father had never gotten to hear. Using a small collection of old photos and voice samples from the late grandfather, the company built a three-dimensional animated likeness that delivered the message: calling Lee’s father “my most precious son,” apologizing for pushing him into childhood farm labor and for opposing his dream of becoming a hairstylist. Though Lee’s father initially said he would not watch the clip, he ultimately sat down to view it, and left the experience in tears.

    Lee’s custom video is far from an isolated case. Across South Korea, an expanding group of digitally fluent consumers are embracing a new AI-powered service that lets people create re-creations of deceased loved ones, giving rise to a fast-growing niche industry populated by local startups. The trend has been amplified by mainstream media, where AI-generated versions of deceased K-pop stars and beloved actors have made broadcast appearances, gradually shifting public acceptance of the technology.

    Vaice, one of the early leaders in this emerging space, currently serves around 300 clients per month, according to CEO Jeongu Won. Most customers are adults in their 40s and 50s seeking AI videos of their own late parents, while others, like Lee, commission videos of grandparents to gift to their still-living parents. A basic three-to-five-minute custom production costs 600,000 South Korean won, equal to roughly $390, and clients almost always write their own scripts. Won says most of these scripts include a simple “I love you,” and many address long-held unresolved conflicts, giving grieving people a chance to close emotional loops they thought were permanently closed. Many families screen the AI videos during annual ancestral memorial rituals or major national holidays, when extended families gather to honor deceased relatives.

    But while many consumers and providers report that the technology brings profound comfort to mourners, it has also sparked intense debate over thorny ethical, psychological and legal questions that existing regulatory frameworks have not caught up to. Experts broadly describe the innovation as a double-edged sword: as AI integrates deeper into daily personal life, it delivers unprecedented cultural experiences that also force society to confront unanticipated ethical shocks.

    Choi Yu Ha, an executive at JL Standard, a company that launched a similar AI recreation service five years ago, notes that early public reception was deeply skeptical. Many bereaved people feared that reopening their grief through a digital simulation would cause more pain than healing. That attitude has shifted dramatically in recent years, however, as simulated appearances of dead celebrities on television have normalized the technology for wider audiences.

    To date, Vaice’s CEO Won says the company has not received any reports from customers that the service worsened their grief. But outside observers warn that the technology carries inherent risks, particularly for vulnerable people who may struggle to distinguish between the virtual simulation and the reality of their loved one’s death.

    Choung Wan, an emeritus law professor at Seoul’s Kyung Hee University, argues that new legislation is urgently needed to protect the dignity and personality rights of deceased people. He says regulations should explicitly ban creating an AI likeness of a person if they explicitly opposed such a use before their death, and set clear boundaries to limit unregulated commercial exploitation of deceased people’s images and voices.

    As the technology advances, the ethical questions are only expected to grow more complex. Many South Korean startups are already experimenting with the next iteration of grief-focused AI: interactive “griefbots” or “deathbots” that allow for two-way real-time conversations between mourners and AI avatars of their dead loved ones. Choung warns that this kind of persistent interaction could do lasting psychological harm. A healthy mourning process, he explains, requires people to accept the permanent absence of the deceased and work through the pain of loss. Simulated ongoing conversation, he argues, undermines that healing process, trapping bereaved families in a fantasy that prevents them from moving forward.

    Even industry leaders are approaching this next step with caution. Vaice’s Won says the company has no plans to launch an interactive chatbot service any time soon, because real-time conversations cannot be pre-vetted or supervised by the company, raising the risk of unforeseen ethical harms.

    Even so, both technological progress and public acceptance are accelerating rapidly. JL Standard’s Choi notes that modern AI can now replicate the fine details of a deceased person’s face — down to individual wrinkles and skin pores — with stunning accuracy, and most customers now report that the AI likenesses are indistinguishable from the real person they remember.

    Yong Man Ro, an AI expert at the Korea Advanced Institute of Science and Technology, who was quoted highlighting the technology’s dual nature, has personal experience with its impact. After his own parents died last year, Ro created a one-minute AI video of their likenesses, which he played for his siblings at a family gathering. When the digital avatars told the assembled family “Don’t worry” and “Take care of yourselves,” everyone in the room was deeply moved. Ro says he and his siblings have never rewatched the clip: one viewing was enough to honor their parents, and they have moved forward with their grief. But for him, the experience illustrates how AI can occupy a gentle, meaningful place in mourning — if it is guided by careful ethical boundaries that society has yet to put in place.

  • Alibaba to pay $600M to settle allegations it allowed illegal drug and equipment sales

    Alibaba to pay $600M to settle allegations it allowed illegal drug and equipment sales

    One of China’s biggest global technology and e-commerce leaders, Alibaba Group, has agreed to a $600 million settlement to resolve a long-running legal dispute with the U.S. government centered on allegations that the company facilitated the import and sale of illegal pharmaceuticals, controlled substances, regulated chemicals, and unapproved pill manufacturing equipment into the United States.

    Alibaba runs two of the world’s largest cross-border e-commerce platforms, Alibaba.com and AliExpress.com, which connect thousands of third-party merchants based around the globe to U.S. consumers. According to U.S. law enforcement allegations, Alibaba’s U.S.-based payment processing arm, AUS Merchant Services, violated federal requirements by failing to put in place sufficient safeguards to block unauthorized merchants from moving unlawful goods into the U.S. through the company’s two major platforms.

    As part of the non-prosecution agreement reached with the U.S. Department of Justice, Alibaba formally acknowledged that between January 2016 and December 2024, it failed to halt approximately 80,000 illegal product transactions that violated the Federal Food, Drug, and Cosmetic Act and multiple other federal statutes. Official documents from the settlement also confirm that internal Alibaba employees had previously flagged gaps in the company’s compliance systems that left the platforms vulnerable to illegal trade. In a number of cases, bad actors even used Alibaba’s built-in messaging tool to redirect buyers to external third-party messaging platforms, where they could complete illicit transactions away from Alibaba’s official monitoring systems.

    The investigation into Alibaba’s practices involved coordinated law enforcement action across multiple federal agencies, including the U.S. Food and Drug Administration, the Federal Deposit Insurance Corporation, and IRS Criminal Investigation. As part of the probe, agents carried out more than 40 undercover purchases that confirmed the presence of the restricted, illegal goods for sale on Alibaba’s platforms. Instead of pursuing criminal prosecution, the Justice Department negotiated a non-prosecution agreement alongside the $600 million settlement, a framework that requires Alibaba to implement sweeping reforms to its U.S.-focused compliance protocols.

    In an official statement released after the deal was finalized, Alibaba noted that the agreement represented a mutually acceptable resolution that will help the company enforce stricter product sale rules for the third-party merchants that operate on its U.S.-facing e-commerce platforms. This settlement marks the latest high-profile regulatory clash between Alibaba and U.S. government authorities, coming after recent moves by the Pentagon to label the company a “Chinese military company” — a designation the firm has formally challenged in federal court, and one that has drawn opposition from Chinese government officials.

    Jarod Koopman, chief of IRS Criminal Investigation, emphasized that the resolution sends a clear message to all global companies operating in the U.S. market. “This outcome underscores IRS Criminal Investigation’s commitment to following the money and ensuring that companies operating in the United States comply fully with federal law,” Koopman said in the official settlement announcement.