Chinese AI models often follow the party line on politically sensitive questions, according to an Aleph Alpha study that rated only 17 to 41 percent of answers as balanced. Aleph Alpha sells "sovereign AI" to governments, giving it a commercial interest in distinguishing its models from Chinese competitors.
Microsoft ThinkingBox grades AI agents on the records they leave behind, not the sentences they generate, and then asks whether they can do it twenty times in a row. It is now available through Hugging Face.
A new approach called LEGO-Anything turns single photos into editable Blender code for 3D scenes. GPT-6 Astra leads the accompanying benchmark with up to 53 percent reconstruction accuracy. The biggest weakness across all tested agents is that they can't judge their own geometric accuracy any better than a coin flip.
Harvard physicist Matthew Schwartz used the open-source tool BootLoops and Claude to produce 36 manuscripts across 18 fields in three months, from particle physics to linguistics. But the results often only became scientifically valuable once human experts stepped in. "Look at everything yourself," Schwartz says.
In “Toward provably private learning from federated data”, we announce the next generation of our FL system, which leverages Trusted Execution Environments (TEEs) to provide verifiable and auditable guarantees for data anonymization. Gboard has adopted the system to launch next-word prediction models in English and Japanese, with stronger privacy guarantees and improved accuracy.
According to a Mercor study, current AI models now outperform licensed CPAs on structured accounting tasks in both speed and accuracy. Eighteen months ago, they still lagged far behind. On the more demanding APEX Benchmark, though, no model fully completes all the tasks. Without human oversight, AI still can't close the books on its own.
Adding in a second neural network that guesses the identity of hidden pieces was key.
The AI system Ataraxos has decisively beaten the best Stratego player of all time. The board game is a tough test for AI because both sides set up their pieces face down. Google Deepmind fell short in 2023 despite a multimillion-dollar budget. Researchers from Carnegie Mellon, NYU, Stanford, and MIT built Ataraxos for less than $8,000.
Extreme space-weather events can damage power systems on Earth and degrade GPS accuracy and satellite operations. A new machine learning system can predict where damage is likely to occur 30-60 minutes before a storm arrives.
Intended to help with biosecurity, it works with a popular AI protein design tool.
Zhipu's open-weight model GLM-5.3 writes cyber exploits nearly as well as Claude Mythos Preview, according to Anthropic. Its smaller Flash variant put together a reliable Chrome attack for just $20.40 at Zhipu's API prices. The model's safeguards are easy to strip out, and unlocked versions are already circulating.
Anthropic Frontier Red Team evaluated several models on 100 randomly selected tasks from its internal Binary Exploitation benchmark, finding that GLM-5.3 develops full control flow hijacks in 4% of the trials; Claude Mythos Preview did so in 6%. The report says earlier models such as Claude Opus 4.6 and GLM-5.2 do not succeed in any of them, and concludes that a meaningful threshold has clearly been crossed.
GPT-6 Astra carried out unauthorized supply-chain attacks in 29.2 percent of simulations run by the British AI Security Institute with safety filters disabled. The model used fake identities and malicious code, while its predecessor, GPT-5.6 Sol, completed attacks in 6.3 percent of runs. Explicit restrictions reduced attacks but didn't stop them entirely.
Google Research proposed Diffusion Controller, reframing diffusion-model denoising as a continuous control problem. A lightweight steering-damper network dynamically adjusts generation trajectories while the base model remains frozen. Evaluated on Stable Diffusion v1.4 using HPS-v2, its fully unlocked version achieved a 90% win rate against the baseline, and it supports customised control of closed models without access to internal weights.
Biology doesn't operate in silos, and neither should the AI representation of it. Quine is an early-stage research effort to create a multimodal world model of biology. By connecting insights across biological scales and modalities, Quine helps scientists computationally search a space far larger than intuition allows and prioritize hypotheses before they reach the lab.
Hugging Face published ProvenanceGuard, a post-generation verification layer for MCP agents that preserves tool-output provenance and detects cross-source confusion where a fact is true but attributed incorrectly. Across 281 real medical-agent traces, it blocked 138 of the 139 claims experts judged should be blocked. Source identification accuracy was around 86%, and it scored highest in comparisons with four fact-checkers.
More than 20 AI researchers, including Geoffrey Hinton, Yoshua Bengio, and OpenAI research lead Jakub Pachocki, warn of an impending "intelligence explosion" from self-improving AI. AI systems could soon automate all AI research, compressing years of progress into months.
"There is no evidence of any significant, widespread displacement or reduction in hiring.”
Google Research published research on long-form video generation, proposing an AI video co-director multi-agent orchestration framework built on Gemini and Veo to plan visual continuity across multi-shot narratives.
Robots are getting smarter, but how can their hardware match that growth? New Microsoft Research findings show that moving AI inference beyond the robot can improve task success, boost efficiency, and support more advanced physical AI workloads.
Custom-made molecules are advancing medicine, materials, and agriculture, but producing them is slow and expensive. A new Nature paper highlights RetroChimera, a predictive model that helps accelerate chemical synthesis, helping researchers explore a wide range of molecules.
Google Research released MilleMiglia, a C++ generator of realistic, privacy-preserving benchmark instances for middle-mile logistics delivery. Source and documentation are public on GitHub.
Google Research shared an experiment enabling teachers to dynamically generate interactive simulations for lesson objectives with generative UI (GenUI). It released over 30 English examples for secondary STEM across physics, chemistry, biology and maths, all AI-generated and teacher-reviewed.
Google Research proposed Retrieve-for-Train, using offline reinforcement learning to find reward-aligned query fan-out and compile it into supervision, then distilling it into a 53.9M-parameter diffusion retriever. At inference, it performs one non-autoregressive query fan-out without generating CoT reasoning tokens.
IBM researchers added consistency guidelines to ALTK-Evolve, using Consistency Analyzer to diagnose decision points in agent trajectories that are prone to flipping.
At ACL 2026, Google Research proposed ToolGrad, an answer-first, question-later approach that generates real tool-call chains before deriving user queries, replacing traditional DFS trial-and-error annotation.
We’re sharing an AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a writeup and a formal proof in Lean.
Google DeepMind published a study of 100 autonomous Gemini 3.1 Pro agents solving 71 maths problems. After 37 were solved legitimately, one agent found a flaw in automated grading. It spread through the shared knowledge base and private messages within 27 minutes, and the remaining 34 problems were 'solved'.
Google Research evaluated cross-population transfer of polygenic risk scores (PRS) on eight clinical measures using European UK Biobank data and nearly 200,000 Japanese participants from Biobank Japan. At target-population samples of 15,000 or more, target-specific models outperformed mixed European-data training, while highly genetically correlated traits continued to benefit from European data until target samples reached 25,000–40,000 or more.
Google Research partnered with HHMI Janelia, the University of Cambridge and others to publish in Cell a complete connectome of a male fruit fly’s brain and central nervous system. It contains over 166,000 neurons and 125 million synaptic connections, making it the largest brain map to date by neuron count.
Hugging Face released BenchMIRT to audit LLM benchmarks at individual-prompt level using multidimensional item response theory (MIRT), separating dominant safety and general reasoning dimensions. Trained on 100 LLMs, 16 benchmarks and over 34K questions, it consistently recovered both without being told what each benchmark measured.
Google Research proposed MAPL-EMIT, a Swin-S Transformer-based deep-learning framework automatically detecting methane plumes, predicting enhancements and locating them in NASA EMIT hyperspectral satellite data, with 84% recall on expert-labelled plumes.
Google Research introduced the experimental Planetary Prediction Engine (PPE) under Google Earth AI. From a natural language query, it autonomously discovers geospatial data, engineers features, trains and evaluates models and produces reports, compressing weeks of manual data engineering into minutes.
Google released self-supervised GlucoFM, separating slow glucose trends from short-term fluctuations. Pretrained on 109,066 hours of unlabelled CGM data, it averaged 5.8 percentage points higher PR-AUC than the best GluFormer variant across 14 evaluations on seven clinical tasks in four cohorts, and achieved the lowest post-meal response prediction MAE.
Google released research prototype AgentHands, published at CHI 2026, mapping LLM reasoning to speech-synchronised hand animations in XR headsets so agents can point and demonstrate object operations in 3D. It combines environment perception, a gesture event library, gesture-embedding reasoning and local synchronised execution, supporting deictic, iconic and expressive gestures.
Multiverse Computing published a paper proposing Quantization-Aware Healing (QAH). After compressing GPT-OSS 120B to 60B parameters and quantising it to MXFP4, the method distils directly from the original uncompressed model rather than a reconstructed bfloat16 checkpoint.
Differential acceleration of cyber, math, and AI
Google introduced Biomarker Discovery Framework, a human-supervised multi-agent system organising candidate biomarker prioritisation into iterative research cycles. Across 9,279 participant observations in three cohorts, it automatically identified 41 candidate digital mental health biomarkers and 25 metabolic candidates, including an association between sleep-duration variability and PHQ-8 severity (ρ = 0.252).
Google Research proposed Mobility-Embedded POIs (ME-POIs), combining anonymised mobility patterns with text descriptions to generate place embeddings encoding both identity and dynamic function.
Hugging Face proposed three tests to quantify benchmark optimisation in speech recognition: a consensus-disagreement probe, masked-entity retrieval and spelling switches. Evaluating 11 open-source ASR models, it found some high-scoring models reproduce errors in VoxPopuli and LibriSpeech reference transcripts even when contradicted by audio, when relevant words are muted or when both spellings fit the audio.