< The research team. From left: Sangyun Chung (KAIST, first author of the MAD study and co-first author of the DNA study); Yong Man Ro (KAIST, corresponding author); Youngjoon Yu (KAIST, co-first author of the DNA study) > Multimodal large language models (MLLMs), which process multiple types of sensory information such as text, images, and audio at the same time, are rapidly expanding the range of applications for artificial intelligence (AI). However, in real-world environments, these models can misinterpret the physical characteristics of sensors, mistakenly identify objects, or claim to hear sounds that are not actually present simply because a certain object appears in a video. These errors are known as hallucinations. A KAIST research team has developed a new technology that corrects such information confusion and physical misperceptions in AI. KAIST (President Choongsik Bae) announced on the 31st of July that a research team led by Professor Yong Man Ro from the School of Electrical Engineering has developed two core technologies that overcome the tendency of existing large language models to rely too heavily on ordinary camera (RGB) images and enable AI to suppress cross-modal hallucinations that occur when different sensory inputs become mixed. < Figure 1. Examples of vision sensor-related questions and responses by recent. It fails to understand the core principle of thermal imaging, incorrectly attributing brightness to reflected sunlight rather than emitted heat. (Courtesy of KAIST) > The first technology developed by the research team is the Diverse Negative Attributes (DNA) optimization method, which helps AI accurately understand the physical characteristics of special camera sensors such as thermal, depth, and X-ray sensors. Existing AI models often failed to understand the physical meaning of such images, for example by mistaking bright areas in thermal images for simple light reflection. The research team built VS-TDX, the first comprehensive
Aug 3, 2026 · via miragenews.com
New guidance explains consent requirements, statutory exceptions and governance expectations The Office of the Australian Information Commissioner (OAIC) has published updated guidance for organisations covered by the Australian Privacy Principles (APPs) that are considering using facial recognition technology (FRT) in high-volume, publicly accessible physical spaces such as retail shopfronts. The revised guidance implements the findings of the Administrative Review Tribunal (ART) in the Bunnings Group Limited matter, which concerned the retailer’s use of facial recognition technology in 62 of its stores between 2018 and 2021. The ART’s findings confirmed that there is a high bar for using facial recognition technology in Australia. In a media release, the OAIC explained that the new guidance makes clear that the Privacy Act neither prohibits nor expressly permits facial recognition technology. Instead, entities must demonstrate that any use complies with the APPs. It confirms that biometric templates and facial images used for automated identification are considered sensitive information and generally attract stronger privacy protections. The OAIC states that entities should adopt a "privacy by design" approach before introducing facial recognition technology. It recommends conducting a privacy impact assessment at the outset of any project to identify, manage and minimise privacy risks, and encourages entities to publish the assessment where possible. Businesses operating facial recognition systems across multiple premises should assess whether each location presents different privacy or security considerations rather than relying on a single blanket assessment. The updated guidance also provides more detailed direction on the lawful collection of biometric information. It says entities must generally obtain an individual's valid consent unless a narrow statutory exception applies. The OAIC emphasises that prominent signage alone will not normally constitute consent, while implied or opt-out consent should rarely be relied upon for the collection of sensitive information. Where businesses seek to rely on an exception
Aug 3, 2026 · via thelawyermag.com
Portuguese cork producer M.A.Silva developed its Nobeltech corks with Bionic Eye, an integrated AI inspection system designed to evaluate each stopper across multiple quality parameters. The technology embeds machine learning into the traditional cork production process to improve consistency, reliability and traceability at scale. Bionic Eye assesses TCA risk, mechanical integrity and sealing performance through imaging and pattern recognition, completing 12 independent inspections before each cork is classified. The system can detect structural cracks, clay contamination, lenticel exposure and insect holes while learning from newly identified defects. Every inspection is recorded, creating a detailed audit trail for each stopper. For winemakers, AI-selected Nobeltech corks offer greater consistency in oxygen transmission rates and long-term ageing performance, reducing bottle-to-bottle variability. By applying machine learning to a natural closure, M.A.Silva combines traditional cork production with measurable quality assurance. AI-Powered Inspection Systems M.A.Silva Introduced Its Bionic Eye Nobeltech Corks Trend Themes - AI Quality Inspection — Machine learning embedded into production lines creates new value in defect prediction, classification accuracy and scalable quality assurance for natural materials. - Traceable Packaging Components — Digital audit trails for individual closures expand transparency across supply chains and support premium product verification in regulated or reputation-sensitive markets. - Smart Natural Materials — Traditional materials enhanced with imaging and pattern recognition can compete with synthetic alternatives through improved consistency, reliability and performance data. Industry Implications - Wine and Spirits — Consistent closure performance and reduced bottle variability strengthen premium ageing claims while supporting more predictable product experiences. - Packaging Technology — Sensor-driven inspection systems introduce differentiated packaging components that combine functional performance with measurable quality records. - Industrial Automation — AI-enabled visual inspection platforms broaden automation opportunities in material-specific manufacturing environments where defects are irregular and evolving.
Aug 3, 2026 · via trendhunter.com
Can generative models also be trained end-to-end? The core turns out to be a for loop. In 2012, AlexNet ended an era with a landslide victory. Before that, image recognition relied on manually designed hierarchical feature extraction pipelines; AlexNet proved a fact that was repeatedly verified later: handing over the entire task to the model for self-learning end-to-end almost always outperforms the carefully designed phased pipelines by humans. From image classification to object detection and then to image segmentation, behind every leap of deep learning is the same logic: let it learn the whole task in one go. There has always been only one exception: generative models. The most powerful and scalable generative models today (whether autoregressive or diffusion models) are not end-to-end. During training, they only learn to predict "one small step", but during inference, they need to unfold this step repeatedly hundreds or thousands of times like a recurrent network. Training and inference do not use the same sampling method. This discrepancy brings up an old problem: the error of each step is fed into the next step, the input gradually drifts away from the distribution seen during training, and errors accumulate layer by layer. Academically, this is called "exposure bias". In other words, "end-to-end performs better" — this core experience of deep learning — has never been truly applied to generative models for more than a decade. Recently, a paper from UIUC and Harvard University attempts to fill this last piece of the puzzle. The authors named this new paradigm Explorative Modeling, abbreviated as XM. Its idea is so simple that it seems almost naive, but it points to a bold conclusion: In addition to parameters and data, generative models actually have a third amplifiable dimension. Project website: https://explorative-modeling.github.io Paper link: https://arxiv.org/abs/2607.27372 Code repository: https://github.com/alexiglad/XM Root of
Aug 3, 2026 · via eu.36kr.com
Space artificial intelligence (AI) corporations TelePIX said on the 3rd that it has completed verification using a space situational awareness camera launched in Jun. last year to estimate a satellite's attitude and compare it with the actual satellite bus's measured values. The space situational awareness camera is a deep-space navigation device co-developed by TelePIX and optical solution corporations LK Samyang. It consists of a wide-field-of-view camera capable of capturing a broad area and an image-processing algorithm, and calculates a satellite's position and attitude based on the locations of stars and planets photographed in space. According to TelePIX, the camera has carried out imaging missions in orbit for more than a year since launch and has continuously secured images sufficient to identify stars. In a recent test, it checked whether the function to estimate changes in the satellite's attitude using only captured images operates normally. As a result of the test, the attitude changes calculated from the star observation images showed a trend similar to the values measured by the satellite bus. The angular velocity, which represents the satellite's rotation speed, also quantitatively matched. The camera identifies individual stars by comparing the positions and brightness of stars captured in the image with a celestial information databases, the "star catalog." It then calculates which direction the camera and the satellite are facing based on the arrangement of the confirmed stars. A test was also conducted to see whether the star recognition function is maintained even in environments where a bright celestial body is captured together. The footage shot in Apr. this year included the moon, but the company said it was still able to continue identifying surrounding stars even when part of the image was saturated by moonlight. This camera applies commercial off-the-shelf (COTS) components rather than parts specially manufactured for exclusive
Aug 3, 2026 · via biz.chosun.com
5 Reds Prospects Earn Major Recognition After Breakout Season In this story: The Cincinnati Reds may have a weaker farm system, ranking 25th in Baseball America's latest ranking, but they had several top prospects contribute to the team's Complex League team advancing to the Arizona Complex League championship series. According to multiple MiLB.com writers, the Reds had five players make their All-Complex League All-Star team: Jesus Colina, Aaron Watson, Jirvin Morillo, Iker Redona, and Steele Hall. Steele Hall Has Been Every Bit As Advertised The Reds selected Steele Hall with their first-round selection in the 2025 draft; he was just 17 years old at the time. He made his professional debut this season, and he answered many questions that were asked about his game. "While Hallâs swing can get a little long at times in games," MLB Pipeline wrote about Hall, "heâs shown in batting practice that he has the ability to be quick to the ball and to drive it gap-to-gap. Heâs a free swinger who will need to refine his approach, particularly in terms of recognizing secondary stuff. The power arrived in his first season. He slashed .288/.387/.504 with 10 home runs, 18 doubles, and his speed was on display with one of his home runs being an inside-the-park home run, two triples, and 25 stolen bases. Hall was promoted at the conclusion of the ACL season to Single-A Daytona, where he went 1-2 with three walks, two RBIs, and a stolen base in his second game. Jirvin Morillo is Ascending Through The Prospect Rankings Morillo debuted this season as the Reds' 29th-ranked prospect. He is now ranked eighth. This season, he slashed .321/.449/.652 with 13 home runs, 13 doubles, five triples, and 53 RBI. He was promoted at the conclusion of the ACL season. Morillo has more
Aug 3, 2026 · via si.com
IAS Residential Fellow Dr Ojwang Alice Achieng delivers a seminar on their research, fully titled "Carbo AI: An Artificial Intelligence-Based Photographic Application to Support Dietary and Lifestyle Education for the Management of Type 2 Diabetes" - Type 2 diabetes is a rapidly growing global health challenge, with dietary management central to its prevention and control. However, estimating the carbohydrate content of meals remains difficult, particularly for locally consumed foods. Carbo AI is an artificial intelligence-powered mobile application that uses photographic image recognition, computer vision, and machine learning to identify foods, estimate portion sizes, calculate carbohydrate content, and provide instant dietary guidance. The project was led by the Human Nutrition and Dietetics Team in the Department of Health and Biomedical Sciences, Technical University of Kenya, in collaboration with experts in computer science, medical physics, and software engineering. Development was supported by Amazon Web Services (AWS) cloud infrastructure and KENET's High-Performance Computing (HPC) platform. Carbo AI aims to improve carbohydrate literacy, strengthen diabetes self-management, enhance nutrition counselling, and demonstrate the potential of interdisciplinary AI-driven innovations to improve health outcomes. Arrivals from 11:45 am for a 12:00 noon start. For those joining in-person, lunch will be served after the seminar from 1:00pm. This event is hybrid format, please use the required booking button at the bottom of the page to choose either in-person or online attendance. (Please note that in-person spaces are limited and booking is required, so we can manage numbers for catering and also the space in the seminar room) By booking a place at this event, attendees agree to behave in a respectful manner such that everyone feels comfortable contributing as they wish. The IAS reserves the right to eject anyone who does not abide by this policy. IAS events are typically recorded, minus any Q&A sessions at the end,
Aug 3, 2026 · via lboro.ac.uk
U.S. Air Force Lt. Col. Melissa Forsyth, left, commander of the 174th Attack Wing Communications Squadron, presents Senior Master Sgt. Jason Hanavan with his promotion order after promoting him during a ceremony at Hancock Field Air National Guard Base, Syracuse, New York, June 8, 2026. Hanavan was promoted to senior master sergeant in recognition of his leadership, professionalism and continued service to the New York Air National Guard. (U.S. Air National Guard photo by Tech. Sgt. Duane Morgan)
| Date Taken: | 06.08.2026 |
| Date Posted: | 08.02.2026 09:04 |
| Photo ID: | 9810044 |
| VIRIN: | 260608-Z-PJ168-1035 |
| Resolution: | 5218x3472 |
| Size: | 3.37 MB |
| Location: | SYRACUSE, NEW YORK, US |
| Web Views: | 4 |
| Downloads: | 0 |
This work, Forsyth promotes Hanavan to senior master sergeant [Image 4 of 4], by TSgt Duane Morgan, identified by DVIDS, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
Aug 2, 2026 · via dvidshub.net
Here’s a new one: smart glasses. “Smart” this, “smart” that. Once “smart” described humans; now, products. The 21st century, awash with smart products, is the smartest century ever. Smart product history is traceable all the way back to 1867: a remotely controlled toy boat. That’s prehistory, really. History proper begins a century later (1968) with Echo IV, a proto home computer that could switch appliances on and off, write a household budget, control home temperature, regulate the kids’ TV access and so on. It was great science but bad marketing: too big, too awkward, it didn’t fit. It was a harbinger, not a hit. The hits of the 21st century were born in the late 20th. Astonishing at first, they are now commonplace. Novelty wears off quickly. The ubiquitous smartphone aside, there are smart cars, smart cards, smart locks, smart doorbells, smart refrigerators, smart saltshakers, smart just about anything. Embryonic smart glasses were Echo IV’s contemporaries, and shared its fate – their time was not yet. It’s now. Beware, warns Spa magazine (July 21-28). They’re dangerous. They range in price from 12,000 to 100,000 yen. The cheapest model displayed on Amazon offers “newest HD camera and speaker, 200 W image stabilization camera glass, video recording, dual microphone, video/ photography, automatic transmission, ENC call noise reduction, audio recording, real-time translation, AI voice assistant, AI image recognition, camera, music player, video recording” – one very smart pair of spectacles! They join the growing ranks of wearable technology – rings, watches, shirts, socks, jackets, trousers – that count, measure and/ or compute steps, heart rate, breathing, sleep, body temperature, foot pressure – everything countable, measurable and computable, and no doubt we’re the healthier for it, though one might venture the suggestion that there are enough things to worry about without all this data
Aug 2, 2026 · via japantoday.com
Default Masking Scheme - Dates2026 - Ongoing - Author - Topics Landscape, Contemporary Issues, Nature & Environment, Fine Art - Location Shanghai, China This series begins with my own photographs of landscapes and examines how automated recognition transforms them into new images. By reworking the masks generated during the recognition process through layering, softening, and recomposition, I create lands I have always wanted to preserve the initial attraction of looking at an image. Before viewing turns into analysis, these works appear simply as seductive, ambiguous, and flattened landscapes. Yet when traced back to their making, that attraction emerges from a working process built upon automated image recognition. In this series, I explore how automated recognition systems interpret landscapes, and how that interpretation reshapes photographic images. Every work begins with my own photographs as source material, alongside masks generated through Camera Raw's automatic landscape recognition during the editing process. My process starts with these automatically generated masks. Rather than preserving the clear boundaries produced by the algorithm, I repeatedly overlay, soften, and reorganise them, transforming the traces of recognition into new image structures. Instead of presenting the software's workflow as an end in itself, I try to retain the images' visual appeal while searching for a balance between intuitive viewing and an awareness of their constructed nature. The red overlay is derived from the default colour used for masking, which itself originates from the Rubylith masking film historically used in printing and plate-making. Once a temporary and largely invisible stage within image production, this colour is reintroduced as part of the final image. It preserves traces of selection while making an otherwise hidden working process visible again, connecting contemporary digital recognition with earlier methods of image production. I hope these works are encountered first as images rather than understood immediately as
Aug 2, 2026 · via phmuseum.com
A wooden box bolted to a platform in a Costa Rican forest can spot a capuchin monkey’s face and choose a puzzle suited to that individual. Get the answer right, and a slice of banana pops out. Nobody has to stand there and run the test. Wild primates are usually studied one of two ways: in controlled laboratory settings or watched loose in the trees with almost no experimental control. Researchers at Emory University and the Georgia Institute of Technology built a new system to close that gap. Marcela Benítez, an assistant professor of anthropology at Emory, led the project, which the team calls CapuchinAI. She has spent years studying monkey behavior both in captivity and in the wild, and she wanted a way to bring lab-grade precision into the forest itself. “The primate brain didn’t evolve in a lab, it evolved in complex, competitive environments,” said Benítez. “Yet primate cognition is rarely studied in the wild because the experimental control needed to measure cognition is difficult in unpredictable environments.” The team first needed a computer that could tell one capuchin from another. They adapted an open-source facial-recognition program called YOLO. Then they trained it on thousands of close-up photos and videos of six adult male capuchins living at the Taboga Forest Reserve. Undergraduates at both universities spent hours drawing boxes around each monkey’s face in old field footage and tagging every image with the correct name. The team tested the finished model on pictures it had never seen, some blurred or lit differently than the training photos. It detected a capuchin’s face in the frame 98 percent of the time, and when it did, it identified the correct monkey 97 percent of the time. Federico Sánchez Vargas, an Emory PhD student in anthropology and the paper’s first author, took
Aug 2, 2026 · via earth.com
AI Rail Station Monitoring Could Move Beyond Passive CCTV 02.08.2026 INTERNATIONAL: AI rail station monitoring could shift security teams from continuous manual viewing towards connected camera, sensor and analytics networks that flag unattended objects and unusual crowd movement in real time. Rail stations combine heavy passenger flows with multiple entrances and open access. That design supports circulation, but it also makes perimeter control difficult and allows a person involved in an incident to blend into a crowd or leave quickly. The pressure is not limited to passenger theft. Security teams may also face antisocial behaviour, attacks on staff, equipment theft and threats to critical infrastructure. In Britain, reported attacks on station staff and police increased by 10% from 2019 despite an 11% reduction in passenger numbers, according to figures reported by the BBC. Connected monitoring replaces isolated camera feeds Conventional CCTV networks often operate as separate systems and depend on staff watching numerous feeds. Older cameras can also deliver images that are difficult to use when teams need to identify people, objects or the sequence of an incident. An IP-based architecture changes the operating model. Cameras can share data with sensors and access-control devices, while a video management system, or VMS, brings feeds and alerts into one interface. Instead of treating every frame equally, analytics can direct attention to events that match configured risk indicators. The applications described in a recent rail station monitoring analysis include unattended-object detection and assessment of crowd behaviour. Software can highlight loitering, an unexpected build-up of people or movement that differs from normal patterns. The same pattern-recognition approach can also support fault reporting and maintenance scheduling. This does not remove the operational role of security staff. It changes the first stage of monitoring: software identifies a possible event, while trained personnel assess the context and
Aug 2, 2026 · via railway.supply
As artificial intelligence becomes more capable, questions about whether machines can think, reason or even replace human intelligence have become increasingly common. Yet decades before AI entered everyday life, B.F. Skinner offered a strikingly different perspective. He wrote, "The real problem is not whether machines think but whether men do." Skinner did not doubt the capacities of the machine; he made us think about ourselves and our ways of thinking. Skinnerâs words suggest that the worst thing is not intelligent machinery but the danger that people will cease to question, analyse and think for themselves. B.F. Skinner (1904-1990) was an American psychologist, behaviourist and writer who made important contributions in the field of learning, behaviour and conditioning that have been influential in the areas of psychology, education and behavioural science. He was always interested in exploring the factors influencing human behaviour through experience and environment. Quote of the day by B.F. Skinner "The real problem is not whether machines think but whether men do." What does this quote mean? This quotation emphasises the fact that what really matters for the progress of society is not whether the machines get smarter, but whether humans keep on using their intelligence effectively. Technology is capable of processing information extremely fast, yet it is unable to substitute human intellect, curiosity, imagination, and moral values. Technology is only as wise as the people who use it Any significant technological revolution comes with tremendous opportunities and new obligations. The computer, Internet, and artificial intelligence are capable of resolving problems and being efficient, yet incapable of choosing between right and wrong or good for society. The decision rests on human evaluation. As Skinner points out, tools gain their meaning with proper judgment. Critical thinking matters more than ever As we live in a world where information is
Aug 2, 2026 · via timesofindia.indiatimes.com
For most of democratic history, the state could see a crowd but not remember it. A police officer watching a rally of 50,000 people could identify only a handful of faces and retain almost none of them by evening. This practical anonymity allowed citizens to participate in public life without fearing that their presence would become a permanent record. Artificial intelligence has begun to dismantle that arrangement. During the recent youth-led demonstrations organised by the Cockroach Janta Party (CJP) at Delhi's Jantar Mantar, the Delhi Police deployed live facial recognition systems mounted on mobile command vehicles. The deployment has since become the subject of a Public Interest Litigation before the Delhi High Court, raising a significant constitutional question: Can the freedoms of assembly and dissent survive when public presence itself becomes data? The controversy is not merely about cameras. Cameras have watched Indian crowds for decades. The fundamental transformation lies elsewhere—it is in the state's newfound ability to remember. A face captured today can potentially become an identity stored, matched and retrieved years later. How does facial recognition technology work? Facial recognition is often described as a smarter camera. That description is misleading. It is a biometric identification system that converts human features into a digital signature. When a camera captures a face, algorithms analyse facial landmarks—the distance between the eyes, the contour of the jaw, the depth of the nose bridge—and transform them into a numerical template known as a ‘faceprint’. The system does not merely record. It interrogates. It continuously asks whether a face corresponds to an identity already present in a database. Connected with police records and surveillance networks, facial recognition can scan thousands of individuals within minutes, without their knowledge or consent. Unlike fingerprints or iris scans, it requires neither cooperation nor physical contact. Presence in
Aug 1, 2026 · via newindianexpress.com
Microsoft's Windows installation files are getting noticeably larger, and the culprit isn't bloatware or legacy code - it's AI. The company's ISO downloads have been steadily creeping upward in size as on-device AI features require increasingly large language models and neural processing capabilities to be baked directly into the operating system. For users with smaller drives or limited bandwidth, this shift is creating real headaches. Microsoft users downloading fresh Windows installations are noticing something unusual - the ISO files keep getting bigger. What used to be a manageable download has ballooned into multi-gigabyte packages that strain both storage and bandwidth, particularly for those working with smaller drives or slower connections. The reason isn't the usual suspects of accumulated updates or legacy compatibility files. Instead, it's the march of AI directly into the operating system. As Microsoft pushes forward with AI-powered features across Windows, those capabilities need to live somewhere. Large language models, neural processing engines, and on-device AI assistants all require substantial storage footprints before they can run locally without constant cloud connectivity. This isn't just about Copilot or voice assistants. Modern Windows builds are increasingly shipping with AI models for everything from image recognition and real-time translation to predictive text and automated system optimization. Each of these features demands its own set of trained models, and those models take up space - sometimes hundreds of megabytes or more per feature. The shift represents a fundamental change in how operating systems are architected. Where previous Windows versions could offload intelligent features to cloud services, the industry is moving toward on-device AI that prioritizes privacy, speed, and offline functionality. But that convenience comes with a storage cost that gets passed directly to users during installation. For enterprise deployments, the implications are particularly significant. IT departments managing fleets of devices need to
Aug 1, 2026 · via techbuzz.ai
Getting your Trinity Audio player ready...
Highlights and lowlights from the week's news
Already have an account? Log In
This article is only available to subscribers
Offer valid for non-subscribers only
Aug 1, 2026 · via chicoer.com
The United States Mission in Nigeria has warned intending travellers against using artificial intelligence or digital editing tools on their passport photographs. The mission said such photos would not be accepted for visa applications. In a statement on its official X handle on Saturday, the mission said passport photos must be recent and must clearly resemble the applicant. It added that submitting an altered image would delay the application process. “Do NOT use AI or digital editing tools on your passport photo. We will not accept photos that are edited or filtered. “Your photo should be recent (taken within the past six months) and look like you (the TSA or CBP agent must be able to tell it’s you). Submitting a digitally enhanced photo will significantly delay your application,” the mission said. The warning was accompanied by a flier reiterating the policy in bold terms. “Do not use AI or digital editing tools on your passport photo. The Department of State will not accept photos that are edited or filtered. “Using a digitally enhanced photo will significantly delay your application,” the flier read. The mission directed applicants to consult the full photo requirements on the State Department’s website, travel.state.gov/photo. The advisory comes amid a broader push by the US authorities to curb the use of AI-manipulated images in official documentation, as facial recognition systems used by border agencies increasingly rely on unaltered, high-fidelity photographs to verify travellers’ identities.
Aug 1, 2026 · via punchng.com
Windows Hello vs. Enhanced Sign‑in Security: Which sign‑in method actually keeps your Windows 11 PC safer, and what's the difference? Windows 11's latest security update expands Enhanced Sign-in Security to external fingerprint readers. Here's what it actually changes. Microsoft is expanding one of the least understood security features in Windows 11. Beginning with the August 2026 update, Enhanced Sign-in Security (ESS) now supports compatible external fingerprint readers, extending the company's most secure Windows Hello experience to devices without built-in biometric hardware. The timing couldn't be better because Enhanced Sign-in Security has confused users ever since Microsoft introduced it. Some users assume it's simply a newer version of Windows Hello, while others think it's reserved for businesses or Copilot+ PCs. The reality is that Windows Hello and Windows Hello Enhanced Sign-in Security use the same sign-in experience, but they protect your biometric data in different ways. After spending time digging through Microsoft's support page, I think that's the distinction most explanations miss. Enhanced Sign-in Security isn't about making facial recognition more accurate or fingerprint sign-in faster. It's about making the entire authentication process harder to attack. Standard Windows Hello is already one of the best security features Before the Enhanced Sign-in Security feature was introduced, Windows Hello had already replaced passwords with a much stronger authentication model. Instead of storing passwords that can be stolen or reused, Windows Hello creates cryptographic credentials that are attached to the Trusted Platform Module (TPM) available on your computer. Facial recognition and fingerprints are used only to unlock those credentials, and your biometric templates remain on the device rather than being uploaded to Microsoft's servers. For the average home user, that already provides excellent protection against phishing, password reuse, and stolen credentials. The Enhanced Sign-in Security feature doesn't replace the architecture already in place. Instead,
Aug 1, 2026 · via windowscentral.com
Gay bars in San Franscisco are installing ID and facial scanners and requiring patrons to give it up if they want to get in, with Cydney Hayes in The San Francisco Gazetteer likening the gates to the TSA and Jacob Ogles in The Advocate pointing out the loss of anonymity and safety in LGBTQ+ spaces. Here's Hayes: Owen had just handed her ID to the bouncer when she was directed to turn and face a small camera before being allowed into the bar. "I was just kind of taken aback," Owen recalled. The camera was hooked up to a monitor and a forensic ID scanner. It looked like the sort of high-security kiosk you'd find while going through airport security. "Why is this at a gay bar, of all places?" At least three Castro bars (Mix, Badlands and Toad Hall) have deployed these PatronScan Guard+ kiosks (previously at Boing Boing), which scan IDs and photograph patrons at the door. The devices record names, birthdays, photos, genders and ZIP codes as well as when you came in. The data is retained for "up to five years" and shared across the network of venues using the system. Bouncers are not required to tell patrons they are being photographed. The Electronic Frontier Foundation questions the legality of the kiosks because they build patron databases that California law forbids. The state's ID privacy statute bars businesses from retaining or using scanned ID data except to verify age or prevent fraud; Joe Mullin wonders why such a serious violation of this law goes unchecked. Californians should be deeply concerned about businesses that collect information from government-issued IDs and use it to build databases about where people go, whom they associate with, and whether they should be allowed into other public gathering places. That concern is
Aug 1, 2026 · via boingboing.net
â Massachusetts Institute of Technology ( MIT ) is reportedly spending more than $3 million to install over 500 AI-powered surveillance cameras across its campus as part of a multi-phase security expansion. While the university says the project is intended to improve campus safety, students have raised concerns over facial recognition capabilities, expanded surveillance and how the collected data could be used. â According to MIT's student newspaper, The Tech, these new cameras will be capable of real-time face and object classification, including detection of motion, crowds, face masks, and other attributes. Defending the rollout, MIT spokesperson Kimberly Allen said the cameras are "part of regular efforts to promote campus security," adding that the data collected is "retained up to 30 days" unless an exception is granted. â More than 500 AI cameras planned across campus â According to documents reviewed by The Tech, installation of the new surveillance system began in November 2025 and is expected to continue through September 2026. â The project includes 521 AI-integrated cameras across academic buildings, residence halls and outdoor locations along Memorial Drive. Of these, 501 are single-lens cameras, and 20 are quad-lens models capable of monitoring multiple directions simultaneously. â The expansion will also add 67 exterior cameras and 13 new "Code Blue" emergency phone towers, each equipped with surveillance cameras, public address systems, emergency lighting and pan-tilt-zoom cameras. â Most of the project's nearly $2 million camera contract has been awarded to Siemens, while telecommunications contractors LCN and Picardi Electric are handling installation, wiring and related infrastructure work. â Cameras use AI for real-time detection â The primary camera supplier is Hanwha Vision, whose Wisenet AI cameras use deep learning algorithms to identify and classify people, vehicles, and other objects in real time. â According to technical specifications cited by The
Aug 1, 2026 · via timesofindia.indiatimes.com