Volume 649 Issue 8099, 29 January 2026

To get the most out of the sensing revolution, different disciplines need to be on the same wavelength. This is a priority for the Nature Portfolio’s newest journal, Nature Sensors.

Dietary change, supported by bold policies, is essential for a sustainable planet.

A supercooled microscopic ferromagnet proves the existence of gyroscopic magnetic behaviour that has been long sought by physicists.

Non-protein-coding genes have been linked to a hereditary condition, retinitis pigmentosa, that causes progressive blindness.

An Austrian cow has shown that some bovines are intelligent enough to employ objects for their own ends.

Chinese factories have embraced the machines, but many activities still require human operators.

The United States is leaving some of the world’s oldest and most influential scientific networks involved in biodiversity research, climate science and conservation. Affected organizations tell Nature that their work continues.

Record-breaking experiment shows that a cluster of thousands of atoms can act like a wave as well as a particle.

Scientists have transformed enigmatic cell structures, called vaults, into storage units for messenger RNA molecules.

A study has outlined eight indicators of toxic masculinity in heterosexual men — and finds that ‘manliness’ is not necessarily a problematic aspect of masculinity.

Learned risk-taking behaviours can persist for years after leaving the lab — and even after taking on a new research topic.

Hijacking the energy-producing organelles from immune cells seems to help tumours in mice to infiltrate lymph nodes.

‘Exercise snacks’ and other forms of everyday movement can greatly reduce the risk of heart disease and death.

For almost two decades, scientists have debated whether sponges or comb jellies are the first animal lineage. Now some are calling for a more harmonious approach.

An exploration of how writing systems, from ancient Chinese characters to modern alphabets, shape language, and whether users of ChatGPT can be said to be authors at all.

Andrew Robinson reviews five of the best science picks.

Actionable goals are needed to guide the world towards what needs to happen most quickly: shifting economies to clean energy sources.

A scientific foundation is required to establish nations’ responsibilities in a hotter ‘overshoot’ world.

A chemist faces a classic early-career dilemma: what should they do next, and how do they start?

Mohammed Amar manages conservation of imperilled flora in Socotra, a Yemeni island known as the Galápagos of the Indian Ocean.

Analyses of hundreds of thousands of papers in the natural sciences reveal a paradox: scientists who use AI tools produce more research but on a more confined set of topics.

Intense droughts have caused water stress and elevated mortality for trees in the Amazon, foreshadowing the future of tropical forests as the climate changes.

Co-optimized design and technology has resulted in cheap, flexible microchips that efficiently run neural-network tasks, as demonstrated in wearable health-care devices.

Conventional benchmarks are becoming less effective at assessing AI performance, but a multi-disciplinary test has set AI systems a fresh challenge.

An in-depth analysis of nutrient availability across mouse tissues reveals their influence on the spread of cancer to other organs.

Genetic analyses of koala pedigrees help to suss out virus-induced genome changes that affect cancer risk and reproductive success.

A simulation approach is used to ascertain the limitations on the structural and chemical information that atomic-resolution electron tomography can determine from noisy electron images.

A quantitative framework for atomic electron tomography enables reliable determination of three-dimensional (3D) atomic coordinates and elemental identities in amorphous materials.

Detection of hot intracluster gas in SPT2349−56 with the Atacama Large Millimeter/submillimeter Array provides new insights into early cluster formation.

Discovery of a protocluster at z = 5.68, merely one billion years after the Big Bang, suggests that large-scale structure must have formed more rapidly in some regions of the early universe than previously thought.

Humanity’s Last Exam, a multi-modal benchmark at the frontier of human knowledge, is designed to be an expert-level closed-ended academic benchmark with broad subject coverage.

Optical control of integer and fractional Chern insulators is demonstrated by circularly polarized optical pumping in tMoTe₂.

Optical spin orientation of itinerant ferromagnets in twisted MoTe₂ homobilayers is demonstrated, enabling control of topological Chern numbers with circularly polarized light.

An integrated optical parametric amplifier on thin-film lithium niobate achieves more than 17 dB gain with less than 200 mW input power.

A flexible digital artificial intelligence chip with a compute-in-memory architecture is fabricated using a commercial low-temperature polycrystalline silicon thin-film transistor process and shows high accuracy in human daily activity monitoring using multimodal physiological signals.

A recently developed class of magneto-sensitive fluorescent proteins are engineered to alter the properties of their response to magnetic fields and radio frequencies, enabling multimodal sensing of biological systems.

An extreme barocaloric effect in NH₄SCN aqueous solutions is enabled by pressure-induced dissolution and precipitation, but without using a separate heat-transfer liquid.

A global atmospheric microplastics dataset is compiled, and comparison with size-aligned microplastic models in the literature indicate that fewer microplastic particles are emitted into the atmosphere than previously thought.

Thirty years of forest demographic data, combined with recent ecophysiological measurements, reveal that intense Amazon droughts sharply increase tree mortality once soil moisture falls below a threshold, and that these hot droughts will become more frequent and intense as Earth warms towards hypertropical conditions.

Metabolic and proteomic profiles derived from fossilized skeletal remains of animals enable inferences regarding physiological health and disease as well as diet to provide reconstructions of ancient soil, vegetation and palaeoclimate characteristics.

AlphaGenome, a deep learning model that inputs 1-Mb DNA sequence to predict functional genomic tracks at single-base resolution across diverse modalities, outperforms existing models in variant effect prediction and enables comprehensive genomic analysis.

WGS data were used from 347,630 individuals with European ancestry in the UK Biobank to obtain high-precision estimates of coding and non-coding rare variant heritability for 34 complex traits and diseases.

Deep learning methods identified a large and evolutionarily conserved X-linked low recombination region in placental mammals that serves as both a barrier to gene flow in hybridizing lineages and an accurate phylogenomic marker.

Artificial intelligence boosts individual scientists’ output, citations and career progression, but collectively narrows research diversity and reduces collaboration, concentrating work in data-rich areas and potentially limiting broader scientific exploration.

How the mouse frontal cortex and striatum interact to implement integrator dynamics controlling movement timing is explored.

The sequential recruitment of a thalamocortical transcriptional cascade enables memory maintenance over long timescales.

Data collected from zoos and aquariums worldwide show that hormonal contraception or permanent surgical sterilization in mammals increase life expectancy, with different mechanisms in males and females.

UBA1-mutant macrophages exposed to inflammatory stimuli undergo aberrant apoptotic and necroptotic cell death mediated by CASP8 and RIPK3–MLKL, offering insights into the mechanisms underlying VEXAS syndrome.

Whole-genome and transcriptome analysis of 1,364 cases of breast cancer from South Korea broadens our understanding of breast cancer biology and reveals genomic features that connect tumour biology with treatment responses and clinical outcomes.

How the complex interplay between multiple nutrients within the microenvironment dictates potential sites of metastatic cancer growth is explored.

Cells can sense mitochondrial damage through signalling via an unimported mitochondrial targeting sequence.

Cas12a3 nucleases constitute a distinct clade of type V CRISPR–Cas bacterial immune systems that preferentially cleave the 3′ tails of tRNAs after recognition of target RNA to induce growth arrest and block phage dissemination.

Rather than adapting to the codon usage of their host, viruses use viral 5′ untranslated regions to initiate translation, which allows them to produce viral proteins in host cells efficiently despite poor codon usage profiles.