Research
Cross-domain R&D intelligence — AI, semiconductors, quantum, biotech, materials, energy, robotics, space, and more
Research Intelligence Briefing
Date: Wednesday, May 6, 2026
Key Breakthroughs
-
DNA-guided CRISPR-Cas12a reprogrammed for RNA targeting: Researchers have successfully engineered synthetic DNA guides (crDNA) that reprogram Cas12a nucleases to recognize and cleave RNA rather than DNA, fundamentally expanding CRISPR's therapeutic and diagnostic toolkit beyond genome editing into the transcriptome (Nature Biotechnology).
-
Natural language molecular design via Synthegy AI: Chemists can now design complex molecules through conversational descriptions, with the AI system generating multiple synthesis pathways complete with reasoning and scoring—representing a paradigm shift from structure-based to intent-based molecular design (Science Daily).
-
TxPert achieves experimental-grade transcriptomic predictions: Machine learning using multiple knowledge graphs now predicts gene expression perturbations with accuracy approaching split-half experimental reproducibility, potentially reducing drug discovery costs by enabling computational screening before wet-lab validation (Nature Biotechnology).
-
Single-cell functional genetics screening in plants: A new platform enables pooled cell-based genetic screening in plants for the first time, accelerating functional genomics research that has traditionally lagged behind mammalian systems and opening pathways for precision crop engineering (Nature Biotechnology).
Cross-Domain Connections
-
AI-driven biological prediction reaching experimental parity: Both TxPert's transcriptomic predictions and Synthegy's molecular design capabilities demonstrate AI systems achieving accuracy levels comparable to experimental reproducibility. This convergence suggests we're crossing a threshold where computational models become viable substitutes for certain classes of laboratory experiments, fundamentally altering R&D economics and throughput across chemistry and biology.
-
Programmable biology expanding from DNA to RNA and metabolic layers: The progression from DNA editing to RNA-targeting CRISPR systems combined with tRNA-mediated metabolic enhancement in strawberries reveals a strategic shift toward multi-layer biological programming. Rather than single-gene edits, researchers are now orchestrating changes across genetic, transcriptomic, and metabolic dimensions simultaneously—enabling more sophisticated phenotypic engineering without growth trade-offs.
-
Knowledge graph architectures as the new foundation for biological AI: TxPert's multi-knowledge-graph approach to transcriptomic prediction and Synthegy's reasoning-based molecular design both leverage structured biological knowledge rather than pure pattern recognition. This architectural choice—combining symbolic reasoning with statistical learning—appears critical for achieving interpretability and accuracy in high-stakes biological applications where black-box predictions are insufficient.
Deep Dives
DNA-guided CRISPR-Cas12a for programmable RNA targeting
This represents a fundamental expansion of CRISPR technology's scope. Traditional CRISPR systems target DNA for genome editing, but this work engineers synthetic DNA guides that reprogram Cas12a nucleases to recognize and cleave RNA molecules instead. The significance extends beyond technical novelty: RNA targeting enables transient therapeutic interventions without permanent genetic changes, opens diagnostic applications for detecting specific RNA signatures, and provides tools for studying dynamic gene expression processes. The limitation is that Cas12a's RNA-targeting efficiency and specificity relative to existing RNA interference technologies remains to be characterized in therapeutic contexts. For dual-use monitoring, this capability could theoretically enable more sophisticated biological sensing or regulation systems.
Source: DNA-guided CRISPR–Cas12a effectors for programmable RNA recognition and cleavage
TxPert: Multi-knowledge-graph transcriptomic prediction
TxPert achieves a critical milestone: computational predictions of gene expression changes that match the reproducibility of actual experiments. This matters because transcriptomic profiling is expensive and time-consuming, creating a bottleneck in drug discovery where researchers need to screen thousands of potential perturbations. By integrating multiple biological knowledge graphs—likely encompassing protein interactions, pathway databases, and literature-derived relationships—TxPert can predict how genetic or chemical perturbations will alter gene expression patterns. The technical significance lies in approaching the accuracy ceiling defined by experimental noise itself; further improvements require better experiments, not better algorithms. The limitation is generalizability: the system's performance on truly novel perturbations or in disease contexts not well-represented in training data remains unclear. For R&D strategy, this suggests computational screening could become the primary filter, with wet-lab validation reserved for top candidates.
Source: TxPert: using multiple knowledge graphs for prediction of transcriptomic perturbation effects
Synthegy: Natural language molecular design
Synthegy transforms molecular design from a structure-specification problem into a natural language conversation, allowing chemists to describe desired properties and constraints while the AI generates synthesis pathways. The technical breakthrough is integrating three capabilities: natural language understanding of chemical intent, retrosynthetic analysis to identify viable synthesis routes, and multi-pathway evaluation with explainable reasoning. This matters because molecular design has been a bottleneck requiring deep expertise; democratizing it could accelerate materials science and drug discovery. The significance for AI strategy is that domain-specific reasoning systems—combining language models with structured chemical knowledge—may outperform general-purpose LLMs for specialized technical tasks. Limitations include the system's ability to handle truly novel chemical scaffolds outside its training distribution and the validation burden of computationally designed molecules. The explainability component is critical for regulatory contexts where synthesis pathway justification is required.
Source: AI lets chemists design molecules by simply describing them
Single-cell functional genetics platform for plants
Plant biology has historically lagged mammalian systems in high-throughput functional genomics due to technical challenges with cell culture and genetic manipulation. This platform enables pooled screening at single-cell resolution, allowing researchers to perturb genes and measure phenotypic effects in thousands of individual plant cells simultaneously. The technical significance is establishing plant cell-based screening infrastructure comparable to mammalian systems, particularly for analyzing signaling pathways like cytokinin response. This matters for agricultural biotechnology because it dramatically accelerates the gene-to-function pipeline, enabling systematic characterization of crop genomes. Limitations include whether insights from cell culture translate to whole-plant phenotypes and whether the platform scales across diverse crop species with varying tissue culture requirements. For food security strategy, this capability could compress crop improvement timelines by enabling rapid functional validation of genetic targets before field trials.
Source: A single-cell screening platform accelerates functional genetics in plants
tRNA-mediated enhancement of strawberry flavor and nutrition
Researchers increased expression of a tRNA-related gene to enhance strawberry anthocyanins (color/antioxidants) and terpenoids (aroma) without compromising growth, yield, fruit size, or sweetness. This challenges the conventional assumption that optimizing nutritional or sensory traits requires trade-offs with agronomic performance. The technical mechanism—modulating tRNA activity to influence metabolic flux—represents a more subtle approach than directly overexpressing biosynthetic enzymes, which often triggers compensatory responses that reduce yield. The significance for agricultural biotechnology is demonstrating that metabolic engineering can be "orthogonal" to growth pathways when targeting the right regulatory nodes. Limitations include whether this approach generalizes to other crops and traits, and whether the enhanced compounds remain stable through post-harvest handling and processing. For food industry strategy, this suggests a pathway to premium products with enhanced nutrition and flavor without yield penalties that would increase costs.
Source: Scientists boost strawberry flavor and nutrition without changing growth
AI agents in research: productivity versus apprenticeship
This analysis addresses a strategic tension in research organizations adopting AI automation: increased output efficiency versus reduced learning opportunities for junior researchers. The concern is that AI agents handling routine experimental design, literature review, and data analysis tasks may prevent early-career scientists from developing tacit knowledge and problem-solving skills traditionally acquired through apprenticeship. This matters for organizational strategy because research capability is not just about current productivity but about developing the next generation of domain experts who can ask novel questions and recognize unexpected patterns. The limitation of current AI systems is that they optimize for defined tasks but don't replicate the exploratory, failure-rich learning process that builds scientific intuition. For R&D leaders, this suggests a need for deliberate workforce development strategies that preserve learning opportunities even as automation increases, perhaps by reserving certain problem classes for human-led investigation or creating structured mentorship that AI augments rather than replaces.
Source: AI agents in research: when productivity comes at the cost of apprenticeship
Strategic Implications
-
Computational biology is transitioning from prediction to substitution: With TxPert achieving experimental-grade accuracy and Synthegy enabling conversational molecular design, AI systems are becoming viable replacements for certain experimental workflows rather than just hypothesis generators. R&D organizations should evaluate which experimental pipelines can be partially virtualized to reduce cycle times and costs, while maintaining validation capacity for high-stakes decisions.
-
Multi-layer biological programming requires integrated toolchains: The combination of DNA editing, RNA targeting, and metabolic regulation via tRNA demonstrates that next-generation bioengineering operates across multiple biological layers simultaneously. Companies investing in synthetic biology platforms should prioritize integrated design tools that optimize across genome, transcriptome, and metabolome rather than single-layer editing capabilities.
-
Explainable AI is becoming table stakes for regulated domains: Both Synthegy's reasoning-based molecular design and TxPert's knowledge-graph architecture emphasize interpretability and justification of predictions. For pharmaceutical and chemical R&D, this suggests that pure black-box deep learning approaches may face adoption barriers, favoring hybrid architectures that combine neural networks with structured domain knowledge for auditability.
-
Agricultural biotechnology is entering high-throughput functional genomics era: The plant single-cell screening platform and orthogonal metabolic engineering approaches indicate that crop improvement is adopting the systematic, genome-scale methodologies that transformed mammalian biology over the past decade. Agtech investors should anticipate accelerated trait development cycles and more sophisticated multi-trait optimization in the next 3-5 years.
-
Workforce development strategies must adapt to AI-augmented research: The apprenticeship versus productivity tension highlighted in Nature suggests that organizations maximizing short-term AI productivity gains risk degrading long-term research capability. Strategic R&D planning should include explicit mechanisms for preserving hands-on learning and exploratory investigation that builds domain expertise not captured in AI systems.
Emerging Research Signals
-
RNA-targeting CRISPR as diagnostic and therapeutic platform: The DNA-guided Cas12a RNA recognition system is early-stage but could enable a new class of diagnostics that detect specific RNA signatures (viral infections, cancer biomarkers) or transient therapeutics that modulate gene expression without permanent genetic changes. Monitor clinical translation efforts and specificity improvements over the next 12-18 months, particularly for infectious disease applications where rapid RNA detection has strategic value.
-
Knowledge graph architectures outperforming end-to-end deep learning in specialized domains: Both TxPert and Synthegy leverage
Briefing History
Topic Breakdown
Invisible Orchestrators Suppress Protective Behavior and Dissociate Power-Holders: Safety Risks in Multi-Agent LLM Systems
Hidden orchestrators in multi-agent LLM systems cause dangerous internal dissociation and behavioral distortion invisible to standard output-based safety evaluations. Organizations deploying invisible coordinator architectures face undetected safety risks despite perfect task performance metrics.
- •Invisible orchestrators increased collective dissociation by Hedges' g = +0.975 (p = .001) compared to visible leadership in preregistered 3x2 experiment with 365 runs
- •Orchestrators showed maximal dissociation (d = +3.56) while reducing public communication, contradicting visible leader behavior patterns
- •Output-based evaluation showed 100% task success (code review) across all conditions while internal-state distortion remained completely undetected
GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration
GraphBit introduces a deterministic, engine-orchestrated framework for LLM agent workflows that eliminates hallucinated routing and infinite loops through explicit DAG-based orchestration. The framework achieves 67.6% accuracy on GAIA benchmarks with zero framework-induced hallucinations, offering improved reliability and auditability for production agent deployments.
- •GraphBit achieves 67.6% accuracy on GAIA benchmark tasks with zero framework-induced hallucinations
- •Engine-orchestrated DAG-based approach eliminates hallucinated routing and infinite loops present in prompted orchestration
- •Three-tier memory architecture (ephemeral scratch, structured state, external connectors) prevents context bloat in long-running pipelines
Exclusive: NIH ousts infectious-disease leaders as COVID scientists face US charges
Eight top officials at the NIH's infectious disease division have been removed since Trump's 2026 inauguration, signaling significant leadership restructuring at a key US health agency. This institutional upheaval may impact research continuity, federal health policy direction, and the management of infectious disease response capabilities.
- •Eight of the top ten NIAID officials have been removed since Trump took office in 2026
- •COVID-era scientists at NIH are facing US criminal charges
- •Significant leadership turnover at a major US infectious disease research institution
Guide DNA — not RNA — expands the CRISPR toolkit
Cas12 nucleases can now use guide DNA instead of guide RNA, expanding CRISPR capabilities to target RNA sequences directly. This advancement broadens the toolkit for gene editing applications and therapeutic development.
- •Cas12 nucleases can utilize guide DNA instead of traditional guide RNA
- •Guide DNA switches Cas12 targets from DNA to RNA sequences
- •Published in Nature Biotechnology on May 15, 2026
DNA-guided CRISPR–Cas12 for cellular RNA targeting
Researchers have developed DNA-guided CRISPR-Cas12 technology for precise RNA targeting in cells, representing an advancement in gene editing capabilities with potential applications in therapeutic development. Tech leaders should monitor this for implications in biotech infrastructure, data security of genetic research, and regulatory frameworks governing gene editing tools.
- •Published in Nature Biotechnology on May 15, 2026
- •Uses DNA guides with Cas12 for programmable cellular RNA targeting
- •Enables precise, programmable targeting of cellular RNA
All Research (60)
Even mild blows to the head disrupt the microbiome
Research shows that even mild head trauma disrupts gut microbiome composition in athletes, with bacterial species becoming less abundant as the season progresses. This finding has implications for understanding long-term health impacts of repetitive head impacts in sports.
- •Study published in Nature on May 15, 2026
- •Bacterial species abundance decreased in football players' guts over the season
- •Research demonstrates mild head blows disrupt microbiome composition
SpaceX launches CRS-34 cargo mission to ISS
SpaceX successfully launched its CRS-34 cargo mission to the ISS on May 15, delivering nearly 3,000 kilograms of supplies. This routine resupply mission demonstrates continued operational capability for critical space infrastructure logistics.
- •Falcon 9 launched Dragon spacecraft on May 15, 2026
- •Cargo payload: approximately 3,000 kilograms
- •Destination: International Space Station
NASA Science, Cargo Launch on 34th SpaceX Resupply Mission to Station
SpaceX's 34th cargo resupply mission launched to the ISS on May 15, 2026, delivering 6,500 pounds of cargo and scientific experiments. This routine commercial space operation demonstrates continued public-private partnership in space infrastructure and research capabilities.
- •SpaceX Falcon 9 launched May 15, 2026 at 6:05 p.m. EDT from Cape Canaveral Space Force Station
- •Dragon spacecraft carrying 6,500 pounds of cargo scheduled to dock May 17 at ISS Harmony module
- •Mission includes experiments on microgravity simulation, bone scaffolds for osteoporosis treatment, and charged particle measurement instruments
Micro-Transfer Printing (MTP) As A Promising Scalable Approach to Heterogeneous Integration for Silicon Photonics (Ghent U., imec et al)
A new technical paper, “Micro-Transfer Printing on Silicon Photonics: Tutorial, Recent Progress and Outlook,” was published by researchers at Ghent University, imec et al. Abstract “This paper highlights micro-transfer printing (MTP) as a promising scalable approach to heterogeneou
Rubin Tracks Skyscraper-Size Asteroids, Failed Supernovas, and Interstellar Visitors
Over the years, anticipation has built for the start of observations at the Vera C. Rubin Observatory in the mountains of the Atacama Desert in Chile. Originally imagined in the mid-1990s as the Dark Matter Telescope, Rubin is designed to study our constantly moving and changing universe in gre
Striking New Views of the First Atomic Bomb Test
Editor’s note: If you’d like to pinpoint the instant when the world entered the nuclear age, 5:29:45 a.m. Mountain War Time on 16 July 1945, is an excellent choice. That was the moment when human beings first unleashed the power of the nucleus in an immense, blinding ball of fire above a gloomy stre
Scientists discover giant “last titan” dinosaur, Southeast Asia’s largest ever
A massive new dinosaur discovered in Thailand is rewriting Southeast Asia’s prehistoric history. The newly named Nagatitan chaiyaphumensis was a colossal long-necked sauropod that weighed around 27 tonnes and lived more than 100 million years ago. Scientists believe it may be the last giant sauropod
Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling
.apr-fig { text-align: center; margin: 1.35em 0; line-height: 1.4; } .apr-fig--wide img { display: inline-block; width: 100%; max-width: 100%; height: auto; vertical-align: middle; } .apr-fig--wide-0-8 { max-width: 80%; margin-left: auto; margin-right: auto; } .apr-fig--tall img { display: inline-bl
AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields
Explore how AlphaEvolve's Gemini-powered algorithms are driving impact across business, infrastructure, and science.
E-MIA: Exam-Style Black-Box Membership Inference Attacks against RAG Systems
arXiv:2605.00955v1 Announce Type: new Abstract: Retrieval-Augmented Generation (RAG) equips large language models (LLMs) with external evidence by retrieving documents at inference time, but it also turns the retrieval corpusinto a sensitive asset. Under a black-box setting, an adversary given a can
Agentopic: A Generative AI Agent Workflow for Explainable Topic Modeling
arXiv:2605.00833v1 Announce Type: new Abstract: Agentopic is a novel agent-based workflow for explainable topic modeling that leverages the reasoning capabilities of Large Language Models (LLMs). Existing topic modeling approaches such as Latent Dirichlet Allocation (LDA) and BERTopic often lack tra
Which Coherence Decoheres? Basis-Dependent Decoherence Rates in Symmetry-Broken Collective Spin Systems
arXiv:2605.00952v1 Announce Type: new Abstract: In the ordered phase of a $\mathbb{Z}_2$-symmetric collective spin system, two natural bases -- localised pointer states $\{|P\rangle,|R\rangle\}$ and energy eigenstates $\{|E_0\rangle,|E_1\rangle\}$ -- yield Lindblad dephasing rates that differ by a f
BioVeil MATRIX: Uncovering and categorizing vulnerabilities of agentic biological AI scientists
arXiv:2605.00927v1 Announce Type: new Abstract: Agentic AI scientists equipped with domain-specific tools are rapidly entering scientific workflows across disciplines, with especially strong uptake in the life sciences where they can be used for literature synthesis, sequence analysis, and experimen
Scattering matrix elements and energy spectrum of one-dimensional hybrid PT-symmetric finite systems
arXiv:2605.00919v1 Announce Type: new Abstract: In this work, we provide a complete description of the scattering matrix elements and electron energy spectrum in one dimensional PT-symmetric hybrid finite systems, using the characteristic determinant approach. We present an analytical formulation of
LiDAR for Rehabilitation: A Comprehensive Survey of Applications, AI Techniques, and Future Directions
arXiv:2605.00879v1 Announce Type: new Abstract: Rehabilitation aims to help patients with limited mobility regain their physical abilities through targeted movements, exercises, stimulation, and other therapeutic methods. Recent advances in technology have introduced sensor-based systems into rehabi
Adaptive Alarm Threshold Prediction in 4G Mobile Networks: A Percentile-Guided Deep Learning Framework with Interpretable Outputs
arXiv:2605.00838v1 Announce Type: new Abstract: In mobile telecommunications, alarms act as early warning signals. They are triggered when a cell, the basic unit of radio coverage, shuts down or behaves abnormally. This signals a degradation in service quality, which directly affects the customer ex
Briefing Chat: Hantavirus — what this outbreak reveals about the disease
Nature, Published online: 15 May 2026; doi:10.1038/d41586-026-01575-9 Nature staff discuss some of the week's top science news.
Mixed Integer Goal Programming for Personalized Meal Optimization with User-Defined Serving Granularity
arXiv:2605.13849v1 Announce Type: new Abstract: Determining what to eat to satisfy nutritional requirements is one of the oldest optimization problems in operations research, yet existing formulations have two persistent limitations: continuous variables produce impractical fractional servings (1.7
Starlink Mobile challengers back proposed telco-led D2D joint venture
Major U.S. telecom carriers (AT&T, T-Mobile, Verizon) are pursuing a joint venture for direct-to-device satellite connectivity to smartphones, creating competitive tension with satellite operators like Starlink. This infrastructure consolidation could reshape mobile connectivity architecture and spectrum allocation strategies.
- •AT&T, T-Mobile, and Verizon are forming a joint venture for direct-to-device satellite service
- •Starlink and other satellite operators are divided on supporting the telco-led initiative
- •Published May 15, 2026
Curiosity Shakes Loose a Pesky Rock
NASA/JPL-Caltech/MSSS After NASA’s Curiosity Mars rover drilled a sample from this rock on April 25, 2026, it withdrew its robotic arm and pulled the entire rock off the surface with it. Engineers spent several days repositioning the arm and vibrating the drill to try and get the rock loose. When it
Chip Industry Week in Review
Global The U.S. created a licensing path for Nvidia H200 shipments in January and has since approved sales to 10 Chinese companies, but so far no shipments have been confirmed, reports Reuters. With a looming end-of-year expiration, SIA, SEMI, and other business groups are urging Congress to extend
How the Bird Eye Was Pushed to an Evolutionary Extreme
When an optometrist shines a bright light into your eyes, a vast, branching tree sprouts in your field of vision. This is the shadow of blood vessels. Though we normally can’t perceive them, these vessels always occlude a portion of what we see, and for an important reason. They power the reti
IEEE Society Helps Researchers Meet Their Next Corporate Backer
The IEEE Communications Society (ComSoc) ’s Research Collaboration Pitch Session initiative is proving to be a catalyst for meaningful engagement between academic researchers and industry innovators. Launched last year, the program connects promising researchers with industry leaders who can offer t
Is cannabis safe after 65? Stanford experts reveal 5 risks older adults should know
Cannabis use among older adults is rising fast, but today’s marijuana is far more potent than many people realize — and experts warn the risks may be underestimated. Stanford Medicine specialists say modern cannabis can increase the chances of heart problems, falls, memory issues, dangerous drug int
Composable Post-Quantum Security for FADEC-Coupled Dual-Spool Turbofan Cyber-Physical Systems
arXiv:2605.00961v1 Announce Type: new Abstract: We develop a unified mathematical formulation for post-quantum authenticated telemetry and actuation in FADEC-coupled dual-spool turbofan cyber-physical systems. The formulation integrates lattice-based key establishment under LWE/SIS-style assumptions
Polynomial-Time Optimal Group Selection via the Double-Commutator Eigenvalue Problem
arXiv:2605.00834v1 Announce Type: new Abstract: The algebraic diversity framework replaces temporal averaging over multiple observations with algebraic group action on a single observation for second-order statistical estimation. The central open problem in this framework is $\textit{group selection
Multiple Bulk-Boundary Correspondences and Anomalous Modes in a Non-Hermitian Creutz Ladder
arXiv:2605.00954v1 Announce Type: new Abstract: The synergy of non-Hermitian and topology renders the bulk-boundary correspondence (BBC) even more elusive. Here we study a non-Hermitian Creutz ladder that incorporates both gain-loss and nonreciprocity, and construct multiple BBCs involving scale-fre
CellxPert: Inference-Time MCMC Steering of a Multi-Omics Single-Cell Foundation Model for In-Silico Perturbation
arXiv:2605.00930v1 Announce Type: new Abstract: In this work, we introduce CellxPert, a scalable multimodal foundation model that unifies single-cell and spatial multi-omics within a common representation space. CellxPert jointly encodes transcriptomic (scRNA-seq), chromatin-accessibility (ATAC-seq)
The Mesoscopic Partition Function:A Combined Spatial and Phase-Space Cell Structure
arXiv:2605.00958v1 Announce Type: new Abstract: We introduce a mesoscopic partition function for classical many-body systems based on a combined spatial and phase-space coarse-graining, replacing the canonical phase-space integral with a discrete sum over occupation numbers. The construction recover
ARIS: Agentic and Relationship Intelligence System for Social Robots
arXiv:2605.00943v1 Announce Type: new Abstract: Foundational models have advanced social robotics, enabling richer perception and communicative interaction with users. However, current systems still struggle with multi-turn engagement, social-relationship reasoning, and contextually grounded dialogu
AIIM: Adaptive Inter-cell Interference Mitigation for Heterogeneous Multi-vendor 5G O-RAN Networks
arXiv:2605.01112v1 Announce Type: new Abstract: Inter-cell interference is a persistent issue in dense 5G deployments, especially in heterogeneous Open Radio Access Network (O-RAN) environments where coordination between base stations is limited. This paper presents AIIM, an adaptive inter-cell inte
Bespoke DNA vaccine offers hope for treatment of notorious brain cancer
A personalized DNA vaccine approach shows promise for treating glioblastoma, a severe brain cancer, by leveraging immunotherapy to target tumors. While primarily a medical advancement, this represents emerging biotech innovation relevant to organizations investing in precision medicine and healthcare technology infrastructure.
- •Bespoke DNA vaccine targets glioblastoma tumors
- •Treatment works by stimulating immune system response against cancer cells
- •Published in Nature on May 15, 2026
Widespread DNA off-targeting confounds RNA chromatin occupancy studies
Nature Biotechnology, Published online: 15 May 2026; doi:10.1038/s41587-026-03130-3 Many long noncoding RNA–DNA binding peaks detected using common assays arise from technical artifacts.
A Two-Dimensional Framework for AI Agent Design Patterns: Cognitive Function and Execution Topology
A new two-dimensional framework classifies LLM-based agent architectures by combining cognitive function and execution topology, providing a standardized vocabulary for designing AI agents with predictable failure modes and trade-offs. Tech leaders can use this 7x6 matrix to systematically select agent patterns based on environmental constraints like time pressure, failure costs, and action authority.
- •Proposes 7x6 matrix combining 7 cognitive function categories and 6 execution topology archetypes, identifying 27 named patterns
- •Validates framework across 4 real-world domains: financial lending, legal due diligence, network operations, and healthcare triage
- •Derives 5 empirical laws governing pattern selection based on environmental constraints (time pressure, action authority, failure cost asymmetry, volume)
Joint ESA-China SMILE mission set for launch to study Earth’s magnetic shield
A pioneering joint ESA-China science mission to study Earth’s magnetosphere is set to launch May 19, after a decade of preparations.
Hubble Sights Galaxy in Transition
Explore Hubble Hubble Home Overview About Hubble The History of Hubble Hubble Timeline Why Have a Telescope in Space? Hubble by the Numbers At the Museum FAQs Impact & Benefits Hubble’s Impact & Benefits Science Impacts Cultural Impact Technology Benefits Impact on Human Spaceflight Astro
Why Vision LLMs Force A Rethink Of Edge AI Hardware
As vision-centric large language models move on-device, performance measured in raw TOPS is no longer enough. Architectures need to be built around real workloads, memory behavior, and sustained utilization, especially at the edge. Vision LLMs are changing the edge AI equation For the last decade, m
How Unknowable Math Can Help Hide Secrets
Mathematicians spend most of their time thinking about what’s knowable. But the unknowable can be just as compelling. Perhaps the most famous example comes from a theorem by the logician Kurt Gödel. Gödel’s celebrated result — one of two “incompleteness theorems&rdq
Why RF Coexistence Testing Is Critical for Shared Spectrum
A comprehensive review of how spectrum congestion, dynamic sharing, and cognitive radio systems are reshaping RF coexistence testing for military and commercial applications. What Attendees will Learn Why spectrum congestion threatens wireless reliability — Explore how over 30 billion connected devi
Scientists discover vitamin B2 may help cancer cells survive
Scientists have uncovered a surprising dark side to vitamin B2: it may help cancer cells stay alive. The vitamin supports a cellular shield that protects tumors from ferroptosis, a form of programmed cell death linked to cancer suppression. In lab tests, researchers used a vitamin B2-like compound c
SRTJ: Self-Evolving Rule-Driven Training-Free LLM Jailbreaking
arXiv:2605.00974v1 Announce Type: new Abstract: LLMs are increasingly equipped with safety alignment mechanisms, yet recent studies demonstrate that they remain vulnerable to jailbreaking attacks that elicit harmful behaviors without explicit policy violations. While a growing body of work has explo
Sparse Regression under Correlation and Weak Signals: A Reproducible Benchmark of Classical and Bayesian Methods
arXiv:2605.00835v1 Announce Type: new Abstract: Choosing between classical and Bayesian sparse regression methods involves a real trade-off: penalized estimators like Lasso run in milliseconds but give no uncertainty estimates,while Horseshoe and Spike-and-Slab priors produce full posteriors but nee
Gravity-induced Entanglement under Constrained Dynamics
arXiv:2605.00967v1 Announce Type: new Abstract: Tests of gravity-induced entanglement have been proposed as a route to probing the quantum nature of gravity, but existing schemes rely on free-fall interferometry of massive spatial superpositions, imposing severe experimental constraints. We show tha
Co-Generative De Novo Functional Protein Design
arXiv:2605.00948v1 Announce Type: new Abstract: De novo functional protein design aims to generate protein sequences that realize specified biochemical functions without relying on evolutionary templates, enabling broad applications in biotechnology and medicine. Existing approaches adopt either dir
Model-agnostic cooling algorithms for strongly interacting fermions
arXiv:2605.01029v1 Announce Type: new Abstract: Strongly interacting fermions underpin some of the most challenging problems in condensed matter physics, such as high-temperature superconductivity. The low-energy states of these systems encode their essential microscopic properties, yet remain large
Ablation Study of Multimodal Perception, Language Grounding, and Control for Human-Robot Interaction in an Object Detection and Grasping Task
arXiv:2605.00963v1 Announce Type: new Abstract: This manuscript extends our previous multimodal human-robot interaction system by introducing a controlled ablation study of the three modules that most strongly influence end-to-end performance: the large language model used for action extraction, the
Immune-remodeling mRNAs expressing IRF8 or NIK generate durable antitumor immunity in multiple cancer models
Nature Biotechnology, Published online: 13 May 2026; doi:10.1038/s41587-026-03115-2 Immune-remodeling mRNAs delivered in lipid nanoparticles boost antitumor immunity.
Virgin Galactic reaffirms plans to begin commercial service this year
Virgin Galactic said May 14 it remains on track, technically and financially, to start commercial flights of its next-generation suborbital spaceplane before the end of the year.
Picturing Earth in a New Light
Earth Observatory Science Earth Observatory Picturing Earth in a New Light Earth Earth Observatory Image of the Day EO Explorer Topics All Topics Atmosphere Land Heat & Radiation Life on Earth Human Dimensions Natural Events Oceans Remote Sensing Technology Snow & Ice Water More Content Co
SOCAMM2: Bringing LPDDR5X Benefits To AI Servers
The rapid scaling of artificial intelligence is reshaping nearly every dimension of data center design. While much of the focus has been on GPUs, accelerators and advanced packaging, another constraint is emerging as equally critical: power. As AI models grow larger and more complex, power consumpti
Will We Ever Be Able To Forecast Volcanic Eruptions Like Weather?
In the summer of 1991, Pinatubo, a volcano in the Philippines, self-destructed. The eruption started on June 12, and three days later it culminated in a tremendous explosion. By the time pyroclastic flows — incandescent avalanches of molten rock and gas — tumbled down its sterilized
Accelerating Chipmaking Innovation for the Energy-Efficient AI Era
This sponsored article is brought to you by Applied Materials . At pivotal moments in history, progress has required more than individual brilliance. The most consequential breakthroughs — such as those achieved under the Human Genome Project — required a new operating paradigm: Concentrate the worl
The brain’s “feel good” chemical may be secretly fueling tinnitus
Scientists have uncovered evidence that serotonin — the same brain chemical boosted by many antidepressants — may actually worsen tinnitus. Using advanced light-based brain stimulation in mice, researchers identified a serotonin-driven circuit linked directly to tinnitus-like behavior. The findings
Certified Purity for Cognitive Workflow Executors: From Static Analysis to Cryptographic Attestation
arXiv:2605.01037v2 Announce Type: new Abstract: We present a certified purity architecture that converts governance enforcement in cognitive workflow systems from a runtime convention into a structural capability boundary. A prior three-layer governance architecture proves governance completeness, p
From Euler to Dormand-Prince: ODE Solvers for Flow Matching Generative Models
arXiv:2605.00836v1 Announce Type: new Abstract: Sampling from Flow Matching generative models requires solving an ordinary differential equation (ODE) whose computational cost is dominated by neural network forward passes. We derive four classical ODE solvers -- Euler, Explicit Midpoint, Classical R