Skip to content
Tech News
Topics: Today This Week This Month This Year

Google DeepMind unveils SynthIDBio watermark for AI-designed proteins

Google DeepMind researchers in London have built a watermarking system, described in Nature, that embeds a hidden statistical signal into both the amino-acid sequence and 3D structure of AI-designed proteins. The tag, called SynthIDBio, is meant to mark proteins as machine-generated without impairing their biological function, allowing them to be identified later in databases.

Sandia's Z machine used to test how Earth's mantle stores water

At Sandia National Laboratories in Albuquerque, mineral physicist Alisha Clark completed the final shot of a multi-year project using the Z machine to recreate the extreme pressures found deep inside Earth. The device, normally used to test the reliability of US nuclear weapons, fired an electrical pulse into water-infused glass samples to examine how water behaves when trapped in Earth's interior during planetary formation.

Study identifies two-step immune mechanism behind autoimmune hair loss in mice

Researchers used genetically engineered mice lacking Treg cells to study autoimmune hair loss, finding that interferon signaling and the AIM2 inflammasome pathway in skin cells and dendritic cells drive hair follicle destruction. Deleting interferon receptors or the Aim2 gene in keratinocytes or cDC1 dendritic cells altered CD8+ T-cell infiltration and hair follicle loss, indicating these pathways are required for disease progression.

Study identifies TBK1 and ULK1 as joint regulators of TFEB nuclear translocation under lysosomal stress

Researchers used TBK1-knockout cells and ULK1 knockdown to examine how cells respond to lysosomal stress induced by monensin, nigericin, salinomycin and MG132. Imaging and biochemical analyses showed that loss of TBK1 or ULK1 altered the nuclear-to-cytoplasmic distribution of the transcription factor TFEB, indicating both proteins are needed for its normal activation.

Google DeepMind adapts SynthID watermarking to AI-designed protein sequences

Researchers have extended Google's SynthID-text watermarking system to ProteinMPNN, a widely used protein sequence design model, creating a method called SynthIDBio-sequence. The approach embeds a detectable signal during the autoregressive sampling of amino acids, using a random seed, scoring function and tournament sampling algorithm similar to the text version. Because protein design typically uses low-temperature sampling that limits randomness, the team also developed a distortionary variant with an added filter to preserve detectability.

Researchers demonstrate self-aligning optical microcomb driven by two octave-spaced lasers

A research team built an integrated frequency-comb device that inverts the usual design: instead of cascading outward from a single pump laser, two lasers spaced exactly one octave apart drive a soliton inside a chip-based resonator, generating a comb that fills the gap between them. The resulting octave-spanning comb, spanning telecom to visible wavelengths, automatically aligns to both pump lasers across multiple fabricated chips and setups, giving it a well-defined zero-frequency offset that is easy to detect and stabilize. The team used this single device to perform optical frequency synthesis, low-noise millimetre-wave generation, and optical clock readout by simply changing which signals the pumps were locked to.

Study models how butterfly wing stripes trick predator motion vision

Researchers built a biologically informed motion-detection model, based on elementary motion detectors used to study cuttlefish hunting displays, to test how butterfly wing patterns affect motion perception. The model simulated bird-like visual limits, including spatial acuity and flicker fusion frequency, at viewing distances scaled to butterfly size. Video recordings of butterflies were filtered to match what a bird predator would actually be able to see before initiating an attack.

Fossil study details anatomy of a late-surviving stem tetrapod

A new paper reports the discovery and analysis of a derived stem tetrapod fossil, describing terrestrial anatomical features and situating the specimen within the broader record of early tetrapod evolution. The reference list draws on decades of research on Devonian and Carboniferous trackways and fossils, including Elginerpeton, Westlothiana, and Polish trackway sites, as well as recent redating work at East Kirkton and new amniote track finds.

DeepMind details Ataraxos, an AI system built for Stratego's hidden-information gameplay

Researchers describe Ataraxos, a decision-making system for the imperfect-information board game Stratego. It combines two interdependent self-play reinforcement learning processes—one for choosing piece set-ups, one for choosing moves—built on transformer networks, alongside a belief network trained on self-play games and a search procedure used at test time.

Study models sampling limits in mouse forebrain lineage-tracing data

Researchers used simulations built from ground-truth clonal lineages to test how partial sampling and varying capture rates affect the ability to reconstruct clonal architecture in the newborn mouse forebrain. They quantified how many recovered multicellular clones are needed to detect all cell types within a lineage and how accurately observed cell-type proportions match the true underlying clone composition.

Study identifies TCR-dependent mechanism shaping γδ T cell tissue surveillance

Researchers used mouse models with tamoxifen-inducible deletion of the Trgc1 gene, which encodes a component of the γδ T cell receptor, to study how these cells patrol tissues like skin and intestine. By selectively removing the receptor gene in specific cell populations and tracking receptor signaling with reporter mice, they examined how TCR expression influences γδ T cell behavior and distribution across tissues.

Today's top topics: openai apple anthropic elon musk iphone duo siri ai homeview apple tv 4k ai agents meta
Browse all topics →  ·  Today's trending topics →