AI News

RSS

AI Moderation Systems Show Striking Inconsistencies In Detecting Hate Speech

This paper by Yphtach Lelkes and Neil Fasching is the first large-scale comparative analysis of seven major AI content moderation systems, evaluating their consistency in detecting hate speech across 1.3 million synthetic sentences. The findings reveal substantial inconsistencies, with some systems flagging identical content as harmful while others deem it acceptable, especially for certain demographic groups.

14 Sep 2025

Rethinking Educational Research For The AI Era: How Scholars Can Rebuild A Broken System

The paper argues that traditional educational research is failing to impact practice and policy due to entrenched methodological flaws, and that the rise of Artificial Intelligence demands a complete epistemological overhaul. It calls for ethical vigilance, methodological pluralism, and innovative, AI-informed approaches to studying education.

11 Sep 2025

Light-Trained Physical Neural Networks Promise Faster Greener AI

Researchers from Politecnico di Milano and international partners developed an in-situ training method for physical neural networks using light-based photonic chips, eliminating the need for digital models. This breakthrough enables faster, more energy-efficient AI computation directly on miniature silicon chips.

11 Sep 2025

New Optical Chip Slashes AI Energy Use With Light-Powered Computing

Researchers at University of Florida have created a silicon chip that uses laser light and microscopic Fresnel lenses to perform convolution operations, a power-intensive step in AI. This optical approach greatly reduces energy use while maintaining near 98% accuracy in machine learning tasks.

11 Sep 2025

AI Learns Evolution’s Logic With Quartet-Based Neural Network

Researchers at Ruhr University Bochum developed a quartet-based deep learning method that embeds phylogenetic tree structures into neural networks. Tested on bacterial 16S rRNA data, it enables AI to extract features aligned with evolutionary relationships rather than random patterns.

11 Sep 2025

AI Pinpoints Immune Cell States Using Force Microscopy Without Labels

A new AI-powered method combines atomic force microscopy with deep learning to rapidly and accurately identify macrophage polarization states without labels or dyes. This approach offers real-time, high-throughput insights into immune function with diagnostic potential across diseases.

11 Sep 2025

AI Can Read Emissions Reports But Can You Trust What It Says?

Researchers created a gold-standard dataset to benchmark AI systems that extract greenhouse gas emissions data from corporate sustainability reports. The study reveals widespread data quality issues, urging caution in using LLMs without verification.

11 Sep 2025

New AI Technique Powers Smarter, Offline Systems Without the Cloud

Kennesaw State University researcher Bobin Deng is developing an AI method that enables powerful models to run directly on personal devices without internet access. By using activation sparsity to predict which data will be needed, his approach reduces energy, memory use, and reliance on cloud computing.

11 Sep 2025

Why Human Hosts Still Outsell AI Streamers in Online Shopping

AI-powered digital streamers are gaining traction in e-commerce livestreams, but they currently underperform compared to human hosts. Real-time interaction and human-like behavior significantly enhance their effectiveness, pointing to hybrid human-AI solutions as the future.

11 Sep 2025

Researchers Warn: AI Could Automate Every Stage of Ransomware

Researchers at NYU Tandon have developed "Ransomware 3.0," a proof-of-concept showing how criminals could use large language models to autonomously execute ransomware attacks. The AI system carried out every phase of an attack in simulations, highlighting future threats and the urgent need for defensive countermeasures.

10 Sep 2025

AI And Satellites Reveal Wildebeest Migration Numbers Are Far Lower Than Believed

For the first time, scientists have used AI and satellite imagery to count the Great Wildebeest Migration, finding fewer than 600,000 animals, less than half the long-standing estimate of 1.3 million. The results highlight new conservation challenges and a breakthrough in wildlife monitoring.

10 Sep 2025

AI Uncovers Hidden Earthquakes And Faults Beneath Italy’s Campi Flegrei Volcano

An AI-powered seismic analysis system has uncovered four times as many earthquakes in Italy’s Campi Flegrei as earlier tools, revealing hidden faults beneath the densely populated volcanic region. The findings sharpen risk assessment, highlighting potential for moderate quakes even without imminent signs of eruption.

10 Sep 2025

AI Shields Power Grid From Cyber And Physical Threats With Low-Cost Devices

Researchers at Sandia National Laboratories have developed a brain-inspired autoencoder AI that detects both physical disruptions and cyberattacks on the power grid in real time. The system runs on inexpensive devices, making it scalable, portable, and suitable for protecting critical infrastructure from storms and hackers alike.

10 Sep 2025

AI Assistant Spots Supernovae With Precision And Slashes Astronomers’ Workload

A new AI-powered Virtual Research Assistant has cut astronomers’ daily workload by 85% by automatically filtering millions of sky alerts and identifying genuine supernova signals. In its first year, it reduced human checks while retaining over 99.9% of real events, paving the way for the data avalanche from the Vera Rubin Observatory.

10 Sep 2025

AI And Drone Imaging Predict Wheat Yields With Unmatched Accuracy

This study developed a transformer-based AI model using UAV multispectral and RGB imaging to predict wheat yields with unprecedented accuracy. The approach allows non-destructive, cost-effective, and early identification of high-yield lines, speeding up breeding for climate-resilient wheat.

9 Sep 2025

Generative AI Designs Thousands of Peptides to Fight Drug-Resistant Bacteria

Researchers at Penn and Duke developed AMP-Diffusion, a generative AI tool that designed 50,000 novel antimicrobial peptides. Two lead candidates successfully treated infections in mice, performing on par with FDA-approved antibiotics and showing no toxicity.

3 Sep 2025

AI-Guided Drones Track Wildfire Smoke to Improve Air Quality Predictions

University of Minnesota researchers have developed AI-powered aerial robots that track wildfire smoke plumes in 3D, enabling real-time, high-resolution data collection. This innovation could vastly improve air quality modeling and early hazard response across a range of environmental threats.

3 Sep 2025

AI System Translates and Edits Human Thoughts Into Images Using Brain Scans

Researchers have developed DreamConnect, a dual-stream AI system that interprets and edits visual mental imagery directly from fMRI signals using natural language prompts. This novel framework allows users to reshape imagined scenes, bridging brain activity and interactive visualization.

3 Sep 2025

AI Reveals Pakistan’s Rivers Face Record Floods And Droughts Every Decade

Researchers used multiple AI models to project the likelihood of record-breaking high- and low-flow events in Pakistan’s major rivers. The findings show basin-specific return periods of 11–15 years, highlighting urgent needs for tailored water management strategies.

3 Sep 2025

UC Riverside Scientists Develop Certified Unlearning Method To Erase Data From AI Models Without Retraining

UC Riverside researchers have created a certified unlearning method that removes sensitive or copyrighted data from AI models without needing access to the original training data. Their framework uses surrogate datasets and calibrated noise to provide provable guarantees while maintaining model performance.

2 Sep 2025