Search

A single portrait reveals the truth | Bella Turkey

Translated title. Original: Tek bir yüz fotoğrafı gerçeği ortaya çıkarıyor | Bella Türkiye

Researchers developed FAHR-Face, an AI system trained on over 40 million faces to estimate biological age and cancer survival from facial regions. The model uses two modules to predict age and survival, remains robust to common imaging variations, and aims to aid clinicians in assessing patient reserve before treatment; broader validation is still needed before clinical use.

The face can reveal more than our age: what artificial intelligence has discovered - Mondosanità

Translated title. Original: Il viso può rivelare più della nostra età: cosa ha scoperto l’intelligenza artificiale - Mondosanità

An AI system called FAHR-Face, using over 40 million photos, links facial features to biological aging and cancer survival signals, with aging cues focusing on certain facial regions and survival cues on others; while promising for assessing frailty, the approach cannot diagnose cancer or predict life expectancy from a selfie, and raises privacy concerns. The study, from Mass General Brigham, was published in The Lancet Digital Health on Oct 19, 2026.

Your name is Sophie and you live in Mortsel': Meta glasses and facial recognition, the dangers for privacy - RTBF Actus

Translated title. Original: 'Vous vous appelez Sophie et vous habitez à Mortsel': lunettes Meta et reconnaissance faciale, les dangers pour la vie privée - RTBF Actus

A collaboration involving ethical hacking and AI is producing a facial recognition system that streams data from glasses to a mobile app, enabling facial tracking and online photo matching of the same person. Experts deem the practice illegal and alarmingly privacy-invasive, likening it to sci-fi scenarios.

Esteban Damian Fuhrmann on X: "Age estimation that runs fully offline on your phone. No cloud. No WiFi 🚫. No YOLO. Photos never leave the device. Live age verification on video, fully on-device: ✅ Big Jeff Simmons (28) → AI 29 😅 Messi (39) → AI 33. Even AI thinks he doesn't age. ✅ Obama (50) →… / X

The document advocates on-device, offline face age estimation that runs rapidly on smartphones, avoiding cloud processing, data leaving the device, and API costs, while enabling real-time, private age verification for many faces per hour.

Face recognition for MacBook – but not with the same security as Face ID - Magazine MACKEN

Translated title. Original: Ansiktsigenkänning till MacBook – men inte med samma säkerhet som Face ID - Magasin MACKEN

Glance adds on-device face recognition to MacBook using Vision/Core ML, but it does not achieve Face ID-level security due to the absence of a depth sensor; it is marketed as a convenience, not a security feature, and is available for free on GitHub and the project site.

New breakthrough in cancer research: AI estimates survival based on facial features | Economx

Translated title. Original: Új áttörés a rákkutatásban: az arcvonások alapján becsüli meg a túlélést az MI | Economx

Researchers debuted AI models that estimate cancer patients’ biological age and survival odds from a single facial photo. The FAHR-FaceAge and FAHR-FaceSurvival systems, trained on tens of millions of images, suggest older-appearing patients have higher mortality and younger-appearing patients lower mortality in a multi-year follow-up.

Face stabilization in AI image generation - Trakya Newspaper

Translated title. Original: Yapay zeka görsel üretiminde yüz sabitleme - Trakya Gazetesi

The document explains techniques for preserving a person’s facial identity in AI-generated images while altering background and other elements, emphasizing explicit prompts and the importance of a clear, high-quality reference face to ensure consistency across scenes.

France Deploys Facial Recognition in Riot Hunt as TAJ Lookups Tripled Under Illegal Use

France employs retrospective facial recognition by cross-referencing social-media images with the national TAJ database to identify riot suspects, a practice permitted under specific legal provisions. The effort, using thousands of cameras and online footage, occurs under strict limits that exclude live public-space scanning and align with EU and national restrictions.

Dutch researchers find clever way to expose AI-generated photos - TW.nl

Translated title. Original: Nederlandse onderzoekers vinden slimme manier om AI-foto's te ontmaskeren - TW.nl

Researchers at the University of Groningen developed a method that checks tiny camera-sensor patterns to help verify an image’s source and detect manipulation. Their startup ForensifAI tested the method with the RDW and is developing products for identity checks, insurance and journalism.

OpenAI launches Control Room for frontier AI oversight

OpenAI announced Control Room, a system that checks research-job authorisation before allowing models to generate output and can block jobs flagged for termination. It aims to address gaps between detecting unsafe behaviour and stopping it, though its effectiveness depends on reliable monitoring and integration.