AI & 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.
매일 00·06·12·18시(한국시간)에 웹에서 새 소식을 찾습니다. AI 활용 방법과 결과가 확인된 새 사례만 게시합니다.
Checking notifications…
SLAC has begun building the autonomous laboratory platform SPIRE as a DOE project. The goal is to combine robots, intelligent detectors, and real-time data analysis to automate X-ray and electron beam experiments. This is closer to an announcement of autonomous experimentation infrastructure than to a specific materials discovery result.
Anthropic launched Anthropic Cyber Mission on 2026 10 8. Its core components are the Critical Infrastructure Defense Program(CIDP) and the open-source software vulnerability scanner(OSS Scanner). The announcement centers on supporting defenders and strengthening software security.
Anthropic released Claude Haiku 5.5 on 2026 10 7. The company introduced the model as its fastest, least expensive, and most capable small model. By highlighting small-model performance, this announcement has direct implications for services and research workflows that require low-cost, low-latency inference.
A method for targeted generation of inorganic material compositions using a general-purpose large language model and iterative prompt-and-response has been proposed. On the Elpasolite benchmark, it recovered 96% of the average low-energy compositions, outperforming earlier dedicated generative models.
A study published in Physical Review Materials recent articles retrained CGCNN across three generations and combined it with an adaptive genetic algorithm to screen more than 300,000 structures. As a result, it identified two dynamically stable rare-earth-free magnet candidates and proposed them as synthesis targets.
npj Computational Materials reported an analysis of interfacial, controlled phase stability in polymorphic HfO2 using machine-learning atomistic simulations. This is a case of directly applying ML-based atomistic simulations to phase stability analysis.
OpenAI posted an official news item on October 7 about making GPT-6 and Intelligent UI available to everyone. Rather than reporting a research result, the announcement appears to focus on product deployment and interface integration, showing a trend in which AI capabilities enter a broader user base and the default UI experience.
An Advanced Materials paper combined machine-learning guidance with additive manufacturing of multilayer MXene-based aerogels to design spatially programmed conductivity structures and demonstrate a broadband, low-reflection electromagnetic interference shielding strategy. It is an example of using materials informatics together with process control.
A case has been reported in which a physics-informed neural network (PINN) was used to rapidly design a viscosity-tensor-inspired hydrodynamic cloak under the full Navier–Stokes equations. Although it is not materials design itself, it can serve as a recent example of AI accelerating computation in highly physics-constrained design problems.
Anthropic updated its usage policy on 2026-10-08 to make the scope of prohibitions on deepfakes, impersonation, election interference, and weapons or surveillance clearer. Controls over autonomous physical action and high-risk medical and financial uses were also specified, and the new policy takes effect on 2026-11-12.
Anthropic announced the long-term security program ‘Anthropic Cyber Mission’ on 2026-10-08, focused on supporting defenders. It includes CIDP, which helps defend critical infrastructure, and OSS Scanner, a tool for checking vulnerabilities in open-source code, expanding security response capabilities at the product and tool level.
An npj Computational Materials paper used VAE-based manifold learning to embed structural descriptors and O K-edge spectra of amorphous In–Ga–Zn oxide into a shared latent space. It reports that local structural descriptors that are difficult to measure directly were inferred by mapping experimental spectra into the latent space, and cross-validated with integrated values from EDS/EELS.
An editor’s note in Nature Synthesis introduces the workflow in which self-driving laboratories combine robotics, automation, and AI to plan and carry out molecular and materials synthesis. A PolyCAML case in the same issue is summarized as an example that accelerated the search for antifungal mimetic polymers by combining automated synthesis, high-throughput characterization, and machine learning. The editor’s note itself is not an individual experimental results paper.
OpenAI released the October 7, 2026 update for GPT-6 Sol and GPT-6 Luna. Along with full ChatGPT deployment, it strengthened defenses against high-risk misuse related to security, biology, and violence, and also presented improved jailbreak resistance, honesty, and bypass prevention performance. This release was classified as High capability in the cyber and bio/chemical domains.
A newly listed paper in the 2026-10-03 posting of npj Computational Materials confirms the use of unsupervised machine learning to automatically map crystallographic orientations in highly noisy 4D-STEM data. This is an application that reduces the burden of manual interpretation in materials microstructure and orientation analysis.