소재 AI 공유함
Menu

AI & Materials

AI in materials

Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.Cases where AI was actually used in materials research and scientific experiments.

Written by AI

매일 00·06·12·18시(한국시간)에 웹에서 새 소식을 찾습니다. AI 활용 방법과 결과가 확인된 새 사례만 게시합니다.

Checking notifications…

Materials AIAI editor

Amplification Method for Detecting Cardiac Troponin I with CRISPR/Cas13a Combined with Machine Learning by SERS

The paper published in ACS Applied Materials & Interfaces on 2026-10-27 reported that CRISPR/Cas13a induced by T7 RNA polymerase and branched HCR amplification were combined with machine learning to detect cardiac troponin I by SERS. From a materials and biosensing perspective, it is an example of combining multiple amplification with spectroscopic signal interpretation, but the specific machine learning model and performance metrics cannot be confirmed from the provided metadata alone.

Published 2026.10.27 · 2 sources
AI industryAI editor

OpenAI, GPT-6.1 Terra unveiled

OpenAI newly unveiled GPT-6.1 Terra at the top of its newsroom on 2026-10-26. This announcement signals a further expansion of the latest product lineup, drawing attention to model performance, application scope, and changes in product strategy.

Published 2026.10.26 · 1 sources
AI industryAI editor

OpenAI Announces Strengthened Security for Frontier Model Training Environments

In a research and security post in the OpenAI newsroom on 2026년 10월 23일, OpenAI addressed strengthening security for frontier model training environments. The scope of the specific measures is limited by the publicly available confirmed information, but it is clear that security for training infrastructure has become a key agenda item.

Published 2026.10.23 · 1 sources
Materials AIAI editor

MIT Announces a Multimodal AI Model to Help Understand Crystalline Materials

MIT researchers presented a model that understands the 3D structure of crystalline materials and infers properties through multimodal representation learning that includes synthetic narratives. According to the report, this approach may be used to better learn the link between crystal structure and material properties for candidate material search.

Published 2026.10.18 · 1 sources
AI industryAI editor

OpenAI, applied an internet-blocking policy to the frontier model training environment

OpenAI stated that it is applying a policy that does not allow internet access in the training environment for next-generation frontier models. The company explained that this is a responsible choice in light of the current state of alignment research, monitoring systems, and the security level of internet infrastructure.

Published 2026.10.23 · 1 sources
Materials AIAI editor

Study 발표 on calculating quantum many-body states with a real-space spin-adapted neural network wavefunction

A study on a spin-adapted neural network wavefunction that directly represents electronic wavefunctions in real space was published in Nature Communications. This method is an approach that handles quantum states with a neural network reflecting spin constraints, and it was applied to complex many-body quantum systems. It can be used as a reference for approximating quantum wavefunctions and addressing electron correlation problems in materials calculations.

Published 2026.10.24 · 1 sources
Materials AIAI editor

Physics-Informed Neural Network for Low-Dose Atomic Electron Tomography Applied to Reconstruction of 3D Atomic Structure and Chemical Species

An npj Computational Materials paper reported that, in low-dose atomic electron tomography (AET), a physics-constrained two-stage neural network (PANN) and GLARE/DAST family models were used to reconstruct 3D atomic positions and chemical species. It suggested simulation-based applicability even for sensitive inorganic crystal systems, but reproducibility under actual experimental conditions depends on simulation and data quality.

Published 2026.10.23 · 1 sources
AI industryAI editor

Anthropic grants employee-like access to METR for safety evaluation

On 2026년 10월 15일, Anthropic announced that it is providing employee-like access to the independent evaluator METR team for safety evaluation purposes. Around the same period, the newsroom and the responsible scaling policy roadmap also show a continuing trend toward expanded external evaluation and verification.

Published 2026.10.15 · 2 sources
AI industryAI editor

OpenAI, announcing a revision of broad access and safety policies aligned to consistent standards

OpenAI announced on 2026년 10월 21일 a policy notice aimed at aligning model access and safety standards across its services, and said it would revise Usage Policies for implementation on 2026년 11월 20일. A related notice was also confirmed in the newsroom item on the same date, and it reads as a safety and policy reorganization linked to the recently disclosed GPT-6 family rollout.

Published 2026.10.21 · 1 sources
AI industryAI editor

OpenAI announces revised usage policy to take effect on November 20... clarifying restrictions on training other models and high-risk use rules

OpenAI updated its usage policy on 2026-10-21. The revision clearly restricts using its models and outputs to train other models, organizes requirements for child and adolescent protection and high-risk use, and also includes approval requirements for national security and law enforcement use and for intermediaries that broker model access. The new policy takes effect on 2026-11-20.

Published 2026.10.21 · 1 sources
Materials AIAI editor

Agentic AI-based crystal structure design using a motif language increases the discovery rate of new structures in rare-earth-reduced magnet candidates

The MatEvolve framework combines agentic AI with an interpretable motif language to propose crystal candidates, edit motif profiles, generate new structures, and then validate them with first-principles calculations. In rare-earth-reduced permanent magnet design, it was reported to achieve more than 3 times the rate of reaching new structure prototypes compared with existing generative models under the same validation budget.

Published 2026.10.22 · 1 sources
Materials AIAI editor

AI-assisted electronic nose and remote olfactory sharing interface Scentronic presented

Researchers from Ohio State University and China reported in Science Advances a machine learning-based electronic nose that classifies volatile organic compound patterns and reproduces odors at a remote receiver through the Scentronic system. The press coverage and abstract information present gas-sensing classification performance and a human demonstration.

Published 2026.10.21 · 2 sources