All posts tagged: detection

Mythos autonomously exploited vulnerabilities that survived 27 years of human review. Security teams need a new detection playbook

Mythos autonomously exploited vulnerabilities that survived 27 years of human review. Security teams need a new detection playbook

Published by skeptic

A 27-year-old bug sat inside OpenBSD’s TCP stack while auditors reviewed the code, fuzzers ran against it, and the operating system earned its reputation as one of the most security-hardened platforms on earth. Two packets could crash any server running it. Finding that bug cost a single Anthropic discovery campaign approximately $20,000. The specific model run that surfaced the flaw cost under $50. Anthropic’s Claude Mythos Preview found it. Autonomously. No human guided the discovery after the initial prompt. The capability jump is not incremental On Firefox 147 exploit writing, Mythos succeeded 181 times versus 2 for Claude Opus 4.6. A 90x improvement in a single generation. SWE-bench Pro: 77.8% versus 53.4%. CyberGym vulnerability reproduction: 83.1% versus 66.6%. Mythos saturated Anthropic’s Cybench CTF at 100%, forcing the red team to shift to real-world zero-day discovery as the only meaningful evaluation left. Then it surfaced thousands of zero-day vulnerabilities across every major operating system and every major browser, many one to two decades old. Anthropic engineers with no formal security training asked Mythos to find remote …

Pennsylvania regulator fines BetMGM over weak fraud detection controls

Pennsylvania regulator fines BetMGM over weak fraud detection controls

Published by skeptic

Pennsylvania gaming regulators have levied a $100,000 fine against BetMGM after concluding the company did not have strong enough safeguards in place to prevent fraud on its online betting platforms. The penalty was approved during the board’s public meeting on March 25 and stems from a consent agreement negotiated with the agency’s Office of Enforcement Counsel following an investigation into BetMGM’s account monitoring and identity verification practices. Fraud rings exploited gaps in identity checks According to regulators, BetMGM’s systems were not strong enough to stop users from opening and operating multiple accounts using stolen or improperly used personal information. Officials pointed to shortcomings in the operator’s Know-Your-Customer (KYC) procedures, which are meant to verify a bettor’s identity before they can place wagers. Investigators said four separate fraud rings exploited those weaknesses over periods ranging from 19 to 34 months. Over that time, the groups created hundreds of accounts using other people’s identifying details and funded them with fraudulently obtained payment methods. operated for approximately 25 months until January 2024 with 1,567 accounts created using personal …

Georgia Could Become the First State With Weapons Detection in All Public Schools

Georgia Could Become the First State With Weapons Detection in All Public Schools

Published by Sam Flores

ATLANTA (AP) — Georgia could become the first state to require every student to be checked for weapons when arriving at a public school each day. “That rifle would have never reached our hallways,” said Daria Lezczynska, a junior at Apalachee High School in Winder, where the shooting took place. “Lives would have been saved. Families would not be grieving, Students like me would not be carrying this trauma.” Some schools have long used metal detectors or required students to carry clear backpacks to cut down on weapons. But a new generation of technology marries computer analysis with cameras or the same electromagnetic fields as metal detectors to detect knives and guns. The systems have spread rapidly through schools, arenas, stadiums and hospitals. “It’s very commonplace for me to walk through a weapons detection system when I enter into a courthouse,” said Chuck Efstration, the bill’s sponsor and Republican house majority leader who represents the Apalachee campus. “Georgia’s students and educators deserve similar security with weapons detection systems inside of every Georgia public school.” There’s …

Outdoor athletes show superior color detection in their peripheral vision

Outdoor athletes show superior color detection in their peripheral vision

Published by José Betancourt

A recent study published in the journal Perception provides evidence that people who play outdoor sports have superior color detection in their peripheral vision compared to indoor athletes and non-athletes. This suggests that intense athletic training in large, open environments can physically shape and improve basic visual skills. The findings indicate that the adult brain retains the ability to adapt its low-level sensory functions based on real-world experiences, long after childhood development has ended. The human eye has a specific biological structure that dictates how well we see certain things. The very center of the visual field is packed with specific light-detecting cells that process bright light and rich colors. As the gaze moves outward toward the edges, a concept scientists call retinal eccentricity, the eyes naturally become less sensitive to color and fine details. Because of this natural biological limit, people usually move their eyes to bring important objects into the direct center of their focus. In fast-paced sports, players cannot always look directly at every moving teammate, opponent, or ball. They rely heavily …

AI could revolutionise glaucoma detection. Here’s 6 things everyone should know

AI could revolutionise glaucoma detection. Here’s 6 things everyone should know

Published by Daniella Wilson

Get the Well Enough newsletter with Harry Bullmore for tips on living a healthier, happier and longer life Get the Well Enough email with Harry Bullmore Get the Well Enough email with Harry Bullmore Artificial intelligence could revolutionise the detection of glaucoma, a common eye condition, while significantly reducing the number of unnecessary specialist referrals, a new study suggests. Published in the Lancet Primary Care, the research from Portugal involved an AI tool analysing eye images from 671 individuals aged between 55 and 65. The technology demonstrated superior accuracy, correctly identifying 78 per cent of people with glaucoma, compared to 75 per cent detected by human doctors. Furthermore, the AI proved more effective at ruling out the condition, accurately excluding 95 per cent of those without glaucoma, against 91 per cent by medical professionals. Crucially, the AI tool recommended just 66 individuals for specialist consultation, leading to 40 glaucoma diagnoses, a stark contrast to the 118 referrals made by eye doctors, which ultimately resulted in the same number of diagnoses. This significant advancement suggests a …

YouTube Adds Politicians and Journalists to Deepfake Detection Tool

YouTube Adds Politicians and Journalists to Deepfake Detection Tool

Published by Daniella Wilson

In a significant move given obvious global events, and with the midterm elections approaching, YouTube is expanding its likeness detection tool to political and civic leaders, as well as journalists, in a bid to curb AI-generated content that may seek to misinform or mislead users of the platform. Politicos and journalists that participate will (after their identities have been verified by YouTube) be able to review videos that have been determined to feature their likeness, and request removal if it violates YouTube’s privacy policies. Generative AI, of course, has made it trivially easy to fake the likeness or voice of someone else. YouTube first announced the tool in December 2024, initially rolling it out to A-list actors and athletes. Last year it expanded it to top creators, and now the company says some 4 million creators in the YouTube Partner Program have signed up to use it. “We’ve always known that there was a need for this tech to go beyond just creators, and so today, we’re excited to announce that we’re going to expand …

AI improves cancer detection in major UK breast screening study

AI improves cancer detection in major UK breast screening study

Published by skeptic

A large UK evaluation finds artificial intelligence (AI) could boost breast cancer detection rates while easing pressure on screening services. AI could significantly improve cancer detection in routine breast screening while reducing pressure on overstretched radiology services, according to a large UK study published today in Nature Cancer. Researchers analysing the use of an AI-assisted system within the National Health Service (NHS) found that integrating the technology into the screening process increased breast cancer detection by more than 10% and reduced clinical workload by over 30%. The findings add new evidence to ongoing discussions about the role of AI in national cancer screening programmes. The evaluation involved more than 10,000 mammograms and examined how AI could support radiologists in identifying potential cancers during routine breast imaging. Large-scale evaluation of AI in NHS breast cancer screening The research was carried out by scientists and clinicians from the University of Aberdeen, NHS Grampian and health technology company Kheiron Medical Technologies, now part of DeepHealth Inc. The work formed part of the GEMINI project, which investigated how AI …

The eyes may be a window into early Alzheimer’s detection, study finds

The eyes may be a window into early Alzheimer’s detection, study finds

Published by Jenni Sidey

Cells along the outer edge of the retina begin to swell and multiply before memory problems appear. That subtle shift, visible only under specialized imaging, may signal the earliest stages of Alzheimer’s disease. Researchers at Houston Methodist report that these peripheral eye changes emerge before major failures occur in the brain’s waste-removal system, suggesting the eye could reveal risk years earlier than current diagnostics. The study, led by biomedical engineer Stephen Wong, Ph.D., examined how retinal support cells behave during the first phase of disease development in mice genetically engineered to develop Alzheimer’s pathology. Findings appear online in the Journal of Alzheimer’s Disease. “The eyes are indeed a window into the brain, but our study reveals that we have been looking at the wrong part of the window,” Wong said. “While most clinical eye exams focus on the central retina, the most critical early indicators of AD appear to be hidden at the periphery of the eye.” Biomedical engineer Stephen Wong, Ph.D. (CREDIT: Houston Methodist) A different part of the eye draws attention Most eye …

AI decision support system for early heart failure risk and detection

AI decision support system for early heart failure risk and detection

Published by skeptic

The STRATIFYHF project aims to develop an AI-based decision support system for the risk stratification, early diagnosis, and management of heart failure, utilising comprehensive data from European clinical centres to enhance patient care and outcomes. Heart failure (HF) is a global pandemic currently affecting up to 15 million people in Europe. It is a complex clinical syndrome associated with impaired heart function, poor quality of life for patients, and high healthcare costs. The STRATIFYHF project addresses this global challenge with an AI-based Decision Support System for HF. The STRATIFYHF aims to develop, validate and implement the first artificial intelligence (AI)-based, Decision Support System (DSS) for assessing and predicting the risk of HF, its early diagnosis, and progression. STRATIFYHF project integrates 1) patient-specific data, i.e. demographic, clinical, genetic, lifestyle and socio-economic, 2) an AI-based digital patient library and algorithms for risk stratification, early diagnosis, and disease progression, and 3) a highly innovative multifunctional AI-based and computational modelling DSS and mobile app for informing a patient-centred, personalised, prevention and treatment strategies (Fig. 1). Fig. 1: Three main …

KRICT microfluidic device simplifies PFAS and pollution detection

KRICT microfluidic device simplifies PFAS and pollution detection

Published by skeptic

A research team in South Korea has unveiled a microfluidic device that could reshape how scientists detect hazardous pollutants in water and food samples. By eliminating the need for filtration and other labour-intensive preparation steps, the technology allows contaminants to be extracted directly from samples containing solid particles, such as sand or food residue. The project was led by Dr Ju Hyeon Kim at the Korea Research Institute of Chemical Technology (KRICT), working closely with Professor Jae Bem You’s team at Chungnam National University. Their newly developed microfluidic device integrates extraction and preparation into a single streamlined process, offering a faster and more reliable approach to environmental analysis. The problem with traditional testing methods Detecting trace contaminants in environmental and food samples is rarely straightforward. Water collected from real-world settings often contains suspended solids, such as soil or organic debris. Before laboratory analysis can begin, technicians typically filter out these particles and then carry out extraction procedures to isolate target compounds. This multi-step workflow presents two major challenges. First, filtration can unintentionally remove trace pollutants …