AI Daily Briefing

Federal disclosures show that Anthropic, OpenAI, and Nvidia each posted their highest-spending lobbying quarter between July and September, with Nvidia leading at $2.2 million, followed by Anthropic at $2.1 million and OpenAI at $1.5 million. The sums are modest next to Microsoft and Google, but signal that the AI firms see policy shaping as increasingly important.

OpenAI's models are now available on all three major cloud provider platforms as the company refines its enterprise strategy.

Gerty, listed on Product Hunt, is a private ChatGPT alternative for Mac that handles its own setup automatically, aiming to give users a private AI chat experience without manual configuration.

HBO Max's new dark comedy The Audacity satirizes Silicon Valley AI titans, from Succession alum Jonathan Glatzer, centering on a half-billion-dollar data analytics CEO desperate to become a billionaire and ascend to 'God Mountain.'

Bloblings is a new Product Hunt project that provides a virtual pet for every Claude Code session.

Equating to the loss of 5% of its 93,000 global staff, job cuts come as bank announces jump in quarterly profits Business live – latest updates Barclays is cutting 5,000 jobs as part of wider cost-cutting plans that its bosses say are being driven by advances in AI. The job cuts will mean losing about 5% of the bank’s 93,000 global staff, with executives confirming they had already cut 1,000 workers since the start of the year. Continue reading...

Tetriz Engg Intelligence + Connectors launched on Product Hunt to link coding sessions with pull requests, issues, quality metrics, and ROI, helping teams measure the real output of AI-assisted development.

Canadian publisher Les Éditions du Boréal has confirmed that best-selling Haitian Canadian author Thélyson Orélien used artificial intelligence to help write his novel C'était ça ou mourir. Orélien had repeatedly denied the allegations, and the book was withdrawn from consideration for France's Prix Goncourt last month.

Discussion | Link

How OpenAI works with distribution partners to uphold consistent safeguards, support responsible access, and clarify expectations in our Usage Policies.

Discussion | Link

Readers react to Simon Jenkins’ article that said it’s more important for children to learn how to speak fluently than study mathematics Simon Jenkins must have had a deeply unpleasant experience of learning maths at school because he seems to have a pathological antipathy towards it (We are teaching children maths they will never use. Isn’t it better to help them speak fluently?, 16 October). I have no qualms with his insistence on the importance of oracy. But his argument breaks down with the claim that it is either oracy or maths, but not both. This is nonsense. Oracy and maths skills should exist side by side. Both contribute to a person’s ability to realise their full potential, and to employment opportunities. Maths drives the understanding of the natural world and contributes vastl…

Simon Willison collects a sardonic remark from Armin Ronacher: AI is fun, but after 30 years of building bad software, the universe is now punishing us for our sins.

Did one of your favorite websites go dark on Wednesday? Some did, as operators took services offline while they waited The post AI found a critical PHP flaw. The bigger failure began in 2024. appeared first on The New Stack.

Investment research firm Citrini Research teamed up with Anthropic to test whether an unreleased Claude model could design scorpion antivenom from scratch. Claude generated hundreds of thousands of candidate proteins, but only 17 bound to any of the targeted toxins, and experts say that is the relatively easy part of a much longer drug development process.

IBM released Granite Speech 5.0 2B, an Apache 2.0 speech recognition and translation model. It ranks #2 on the Open ASR Leaderboard with 5.68% average English WER, alongside updated Turbo CTC and 4.1 2B Plus models.

NVIDIA has expanded its Cosmos platform with new training tools, simulation checkpoints, and Cosmos Data Factory blueprints to help developers build world models for physical AI. The platform ranks top five on six open model leaderboards, and new tools span data generation, model training and optimization, and simulation evaluation.

The paper introduces T-WLA, a world-language-action architecture that uses shared visual memory to drive both base action generation and lightweight tactile correction, letting robots react to contact changes without rerunning the slower visual policy. It reduces finger-joint MPJPE by 30% against the strongest baseline and reaches roughly 50.0% average success across four real-world dexterous manipulation tasks.

arXiv:2610.28791v1 Announce Type: new Abstract: Target-based extrinsic calibration estimates the rigid transform between sensors from a few observations of a known target, and its result is commonly assessed by the root-mean-square error of the fitted residuals. However, this scalar error is computed on the same observations and therefore cannot indicate whether to accept the result, re-extract, recollect or add observations. This paper presents VeriCalib, which formulates extrinsic calibration as a verifiable estimation problem and reports a calibration state. The key insight is a two-level extrinsic compensation mechanism, in which the extrinsic compensates the rigid part of each observation error at the residual level, and rotation and translation compensate each other at the parameter…

This paper introduces a framework for physical human-robot interaction that pairs a robust wrench observer with a hybrid pose and force controller. A neural network compensates for friction and unmodeled dynamics, task-null-space decomposition limits the influence of human disturbance on the observer, and the hybrid controller tracks end-effector pose and force without force/torque sensor measurements.

arXiv:2610.28701v1 Announce Type: new Abstract: Accurate ground plane normal estimation is essential for odometry in spherical mobile mapping systems (SMMS), which face challenges from rapid rotations and limited ground coverage. We propose two RANSAC-based approaches: single-scan estimation with normal vector scoring and multi-step validation, and LiDAR-inertial odometry (LIO)-based estimation using a global 3D map. Evaluated across diverse real-world datasets against terrestrial laser scanner ground truth, results show that LIO-based estimation significantly outperforms single-scan methods in accuracy and robustness. Integrating ground normal estimates into the motion model improves pose estimation, supporting autonomous navigation in challenging and hazardous environments.

arXiv:2610.28835v1 Announce Type: new Abstract: Conventional visual sensing faces a fundamental dilemma: information that should remain private today may need to be recovered tomorrow. This challenge is becoming increasingly important as cameras become pervasive in our daily surroundings. Existing safeguards either protect visual content only after an image is formed or suppress information during capture in ways that prevent its later recovery. Here, we show through an information-theoretic analysis that privacy and recoverability cannot be achieved simultaneously in static optical systems. Information preserved in a measurement for later reconstruction also remains available to an observer of that measurement, limiting the privacy such systems can provide. To overcome this limitation, w…

arXiv:2610.28827v1 Announce Type: new Abstract: We present WorldStream, an open-ended 3D world model with three simultaneous capabilities: it extends playable terrain beyond any pre-specified map, acts on open-ended language in the running world, and restores visited places with their edits. A hierarchical planner over a typed decision model emits structured decision state: Macro fixes settlement, ecology, transport, and hydrological structure across a group of tiles; Micro specializes that contract into tile semantics under measured terrain and accepted neighbor commitments. Deterministic local code compiles accepted plans into terrain, roads, settlements, vegetation, and entities. Streaming separates their reach: plans run ahead along the predicted corridor, compilation only where movem…

The arXiv paper SAIL introduces a differentiable framework that generalises classical spatial autocorrelation statistics such as Moran's I and Geary's C into a learnable operator on graphs. By jointly learning feature projections and inter-node neighbourhood affinities, it produces interpretable local and global spatial coherence descriptors for supervised prediction and unsupervised hotspot discovery.

A WACV 2027 paper performs a controlled audit of the MS1MV3 face recognition training set, identifying five categories of residual data defects and showing through IResNet-50/ArcFace training that targeted fixes can maintain or even improve recognition performance while requiring fewer optimization steps, with duplicate removal yielding the clearest benefit.

A new arXiv paper introduces the Motion Equivariant Convolutional LSTM (MELSTM), which treats motion in sequential data as a gauge symmetry and uses a learned velocity model as the gauge field, outperforming a convolutional LSTM and a flow equivariant counterpart at equal parameter count.

The paper introduces SimHack, a simulator that learns reward hacking training dynamics from a handful of smaller-scale reference RL runs and predicts, before an actual RL run, when detectable reward hacking becomes dominant under a target recipe, enabling lower-risk recipe selection.

The paper proposes simultaneous neural topology optimization (SNTO), a framework in which state, adjoint, and design variables are all represented by custom neural networks trained jointly in a single optimization loop, enabling stress-constrained topology optimization at millions of degrees of freedom.

This survey by Hamza Kheddar reviews more than 100 studies on implementing reinforcement learning on FPGA and integrated-circuit platforms, organizing the literature by RL family, abstraction level, hardware realization strategy, and application domain, while comparing resource use, latency, throughput, power, and performance gains and outlining recurring architectural patterns, practical limits, and open research directions.

A conference preprint on arXiv introduces TF-IG, a dual-branch, interaction-gated time-frequency learning framework for PPG-based heart rate estimation. With only 84.6K parameters, the model reaches a mean absolute error of 1.227 BPM and a Pearson correlation of 0.9934 on the BIDMC dataset, illustrating a practical accuracy-complexity trade-off for lightweight wearable settings.

A rush of new data centers could bring air pollution roaring back to levels not seen since the 2000s in Texas, which has become ground zero for the physical infrastructure behind generative AI. New and proposed data centers are demanding a tremendous amount of electricity. How developers and utilities meet that demand - with gas turbines or renewable energy - will determine just how dirty things will get. So far, though, things are looking grim. Tech companies and data center developers are devising new ways to get enough electricity, including building out their own power sources. But if all planned data center projects were to plug into … Read the full story at The Verge.

PHP powers roughly 70% of the world’s websites, according to Wiz Research, so a flaw in its core engine affects The post AI found flaws in the code behind 70% of the web. The fix took two months. appeared first on The New Stack.

U.S. Hunger, a nonprofit addressing food insecurity, uses Amazon Quick to connect organizational data with leadership decisions. Leaders explore applicant data directly, share KPI definitions and financial context in Quick Spaces, and investigate policy questions with Quick Research.

For decades, a fundamental question about the difficulty of computing in high-dimensional spaces has resisted proof. New research resolves this question, removing a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing. The Kannan–Lovász–Simonovits (KLS) conjecture was proposed in 1995. It asks whether, for a broad class of high-dimensional […] The post Solving a 31-year-old grand challenge in mathematics – the KLS conjecture appeared first on Microsoft Research.

Dr. Christopher Stubbs, an experimental physicist and former Dean of Science at Harvard University, has joined Anthropic’s Long-Term Benefit Trust. He brings to the Trust decades of scientific expertise across observati…

NVIDIA has launched Minerva, an AIOps platform that unifies telemetry from infrastructure and AI workloads and applies purpose-built AI models to surface timely, actionable insights, helping teams detect problems earlier and keep AI factories productive. IREN is among the early users, with Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI and Tesla also planning deployments and Dell Technologies, Lenovo and Supermicro offering it in their AI factory operations portfolios.

This tutorial explains how to build a three-agent "night shift" team in roughly 150 lines of Python without any framework: Scout researches, Quill writes, and Red edits. The agents share one LLM engine, hand work off through files rather than conversation, and are constrained by code-enforced guardrails and a human approval desk so that nothing goes public without a manager's sign-off.

This essay argues that AI environments are becoming a core ingredient of model capability, not merely benchmarks. It traces their evolution from Atari, OpenAI Gym, and AlphaZero through MuJoCo to the RLVR-driven post-training behind Tulu 3 and DeepSeek-R1, surveys a new generation of agent environments including SWE-Gym, WebArena, OSWorld, TheAgentCompany, Tau2-bench, and AgentMercury, and lays out what makes an environment effective: outcome-aligned rewards that resist hacking, causal fidelity, affordable and repeatable practice, structured difficulty, and improvement that transfers beyond the training world.

Security researchers disclosed two PHP-engine memory-corruption vulnerabilities, dubbed PHP2Shell (CVE-2026-109618 and CVE-2026-109619), enabling unauthenticated remote code execution and Heartbleed-style memory disclosure. OpenAI says its Calif researchers found the flaws using GPT-5.6 Sol under the Patch the Planet program, while secondary reporting says Wiz independently discovered the same bugs. A roughly two-month coordinated remediation involving the PHP Foundation, Cloudflare and governments preceded disclosure, with Cloudflare deploying WAF mitigations before the patch. The piece stresses that this is a multi-party discovery and remediation story and cautions against treating leaked-detail timelines, server-count figures and outage reports as independently verified.

At GTC Berlin, European banks, payment companies and financial data providers showcased production-scale AI built on NVIDIA accelerated computing, open weight models and their own infrastructure. Revolut is rebuilding its stack on the PRAGMA foundation model, while Starling Bank, Lloyds, bunq, NatWest, Santander, LSEG, HSBC and Stripe reported gains in sovereign cloud training, transaction modeling, agentic assistants, speech recognition and fraud detection.

NVIDIA showcased its Earth-2 open AI model family at GTC Berlin, including StormScope EU, a regional short-term forecasting model trained on European weather and climate data. Organizations including AXA, Vaisala Xweather, the University of Manchester, TotalEnergies and Space42 are applying these models to cyclone risk, lightning-enhanced nowcasting, air-quality modeling, solar forecasting and radar-based terrain estimation.

A group of AI infrastructure companies including Nvidia, Microsoft and Cisco has told UK ministers they have roughly 12 months to defuse public opposition to datacentres or risk US-style protests that could hold up construction. The submission points to successive dry summers and rising energy bills as factors sharpening scrutiny of the large computing facilities used to train and run AI models.

Perplexity and NVIDIA researchers propose JOLT (Joint On-policy Learning and Teaching), a post-training method in which a single model plays both teacher and student, with the teacher itself trained via outcome rewards and a KL penalty toward the student. On MATH it matched GRPO's accuracy using roughly 6.5x fewer processed training tokens, and JOLT+ posted the best mean score in all five tested settings, including a 9.9-point gain over GRPO on AppWorld.

This reproducible tutorial benchmarks Cactus Compute's 16.9 MB CPU speech-to-text model, Whistle, against Whisper tiny and base on seeded LibriSpeech test-clean and test-other samples from an Apple M4. Using paired speaker-cluster bootstrap intervals with Holm correction, the author finds Whistle leads over greedy Whisper base on test-other but shows no significant gap against beam search, and its speed advantage narrows as clips get longer. The guide also probes a 30-second single-pass window where the pinned engine silently drops the middle of longer audio, plus streaming latency, timestamps, keyword biasing, other languages, noise, embeddings, and speech-to-tool calls.

The paper introduces DAFIC, a geometry-guided framework that brings monocular depth into few-shot image classification through RGB-D self-supervised pre-training, a Depth-guided Geometric Adaptation module, and Spatial Bidirectional Patch Matching to counter background clutter and RGB appearance ambiguity. Accepted at WACV 2027, it reports strong results across four benchmarks.

arXiv:2610.28843v1 Announce Type: new Abstract: Visual question answering requires models to assign attributes and relations to the correct objects, a problem known as visual binding. Structured prompts make these assignments explicit through labeled fields. Yet an improvement from the full prompt does not reveal which labels help or how they interact. First, we isolate field-label effects across ten models and six tasks by holding the image and claim fixed. These comparisons reveal differences that overall accuracy can hide. With a relation label present, adding object-role labels helps Qwen3 but hurts Qwen2 on Spatial. Second, we compare auxiliary hints that preserve or swap the question's role assignments against a no-hint baseline. Across two image sets, Qwen3's advantage with preserv…

arXiv:2610.28829v1 Announce Type: new Abstract: Instruction-guided video generative models offer a scalable solution for stress-testing autonomous driving systems by simulating diverse environmental conditions. However, driving scenes operate within a dense Operational Design Domain (ODD) governed by strict traffic semantics. When applied to global driving-condition edits, general-purpose instruction-guided video editors frequently corrupt these semantics, leading to structural hallucinations and ``temporal shock.'' This work distills and protects traffic semantics during video editing at three levels. First, we construct a hybrid dataset by combining temporally aligned real-world videos with synthetic edits, balancing foundational scene structure with unambiguous appearance signals. Seco…

A NeurIPS 2026 paper introduces Register Anything Model (RAM), a unified point cloud registration model trained on indoor, outdoor, and object data that robustly estimates 6-DoF transformations across sensors and environments. Its key ingredient is a geometry-aware normalization strategy that adaptively voxelizes and normalizes point clouds into a standardized geometric space, paired with a hierarchical coarse-to-fine pipeline and an efficient geometric transformer. Across 12 benchmarks it achieves state-of-the-art inlier ratio and registration recall, outperforming models trained separately per domain.

The paper introduces JUICE, a single flow-matching Transformer that models the joint distribution of images and text directly inside frozen pretrained encoder spaces, giving each modality its own noise level so that which modalities are clean and which are generated determines the task. At 2.1B parameters and 224x224 resolution with no language-model initialization, it reports up to 0.88 on GenEval and 73.1 on VQAv2, and it substantially outperforms independent sampling on maze tasks that require image and text to agree.

arXiv:2610.28840v1 Announce Type: new Abstract: Early-grade reading screening is critical for identifying children at risk of reading difficulty. Existing AI-based screeners are often limited in the scope of their analyses and evaluation metrics, or are proprietary systems that lack transparent model specifications and evaluation. We present KIRA, a dual-pathway screening system that pairs word-level fluency scoring with detailed phoneme-level error detection. We adapt pretrained word and phoneme recognizers to children's speech, reducing word error rate from 13.88% to 8.84% and phoneme error rate from 41.66% to 23.00%. Powered by these child-adapted models, we implement Word Reading (WRE) and Oral Reading Fluency (ORF), two foundational tasks in K-2 screening. The system integrates these…

A new paper examines how to explicitly exploit parallel data structure during multilingual pretraining, finding that parallel pretraining (PPT) alone does not consistently improve language understanding, but a two-stage approach—PPT to build a cross-lingual foundation followed by standard pretraining to adapt it—consistently beats standard pretraining in both high- and low-resource settings.

This paper investigates whether large language models form internal representations corresponding to all 26 CATO factors used in case-based legal reasoning. Experiments on three open models—Qwen3-1.7B, Qwen3-8B, and Gemma-2 2B—examine linear decodability and sparse localization with sparse autoencoders, finding that all 26 factors are linearly decodable with best-layer test AUCs from 0.86 to 1.00, and reporting 78 model-specific layer-feature coordinates.

The paper argues that existing evaluations of collusion among LLM pricing agents focus on outcomes while missing enforcement, which is what actually defines a cartel. The authors introduce an audit that forces a member to cheat in counterfactual copies of a market, and use it to examine what reinforcement learning teaches agents in pricing and trading.

arXiv:2610.28772v1 Announce Type: new Abstract: Psychotherapy fidelity assessment measures how faithfully a therapist adheres to a treatment protocol, a critical determinant of positive patient outcomes. However, fidelity assessment is bottlenecked by manual expert review, rendering the process labor-intensive and unscalable. To scale therapy supervision and support clinician self-learning, we present an integrated application of large language models (LLMs) that automates psychotherapy fidelity workflows and generates actionable supervisory feedback from authentic, long-form clinical transcripts. We evaluate three open-weight models across multiple task decomposition configurations and reasoning strategies, demonstrating that evidence-first reasoning improves fidelity estimation. Further…

Researchers propose SkillEval, an evaluation framework that uses LLMs' meta-cognitive abilities to automatically discover interpretable skills in benchmark questions and applies psychometric cognitive diagnosis models to estimate per-skill abilities, replacing the usual aggregation of all items into a single scalar score.

An arXiv paper examines how well the 305M-parameter audio encoder shared by Gemma 4 and EmbeddingGemma 2 performs on Saudi and Gulf Arabic speech recognition without altering any of its weights. A frozen system trains only a 22.6M-parameter CTC reader, while an adapted system adds low-rank adapters, a hybrid CTC/attention reader and joint decoding with a sub-word language model. The adapted system is statistically indistinguishable from the 2.07B-parameter Cohere Transcribe Arabic in WER on several public tests, is significantly better on telephone-coded Emirati speech and is faster on some setups, while still trailing on long-form Saudi speech and ranking 4th of 15 systems on a blind call-centre sample.

This paper introduces ReAPO, a simple modification to GRPO derived from viewing credit assignment through the lens of Stochastic Shortest Paths. ReAPO allocates additional generation to failed trajectories so that successful retries are reinforced without rewarding preceding failures, assigning credit to each attempt in a theoretically principled way. Across Qwen3 and OLMo-2 models from 0.6B to 4B and eight mathematical and scientific reasoning benchmarks, ReAPO improves first-attempt performance over GRPO and recent baselines; on Qwen3-0.6B it matches the peak accuracy of GRPO with twice the group size or twice the training steps using 39% and 31% fewer generated tokens, respectively.

A new arXiv paper compares the GraphLLM node-classification results reported by three recent studies against inexpensive baselines tuned with validation-only selection, finding median gaps of just 2.3 to 2.8 percentage points. The author also documents several evaluation risks in the released code and papers, including test-set hyperparameter selection, test-pool labels leaking into training, undisclosed reuse of baseline numbers, and a text-label misalignment in a Cora loader.

The paper introduces STAR-XPC, a Set Transformer model that maps the molecular constituents, mass fractions, and four bulk physical descriptors of polymer composites directly to the X-ray transmission ratio I(t)/I0 in 1-3 milliseconds, approaching Monte Carlo accuracy and offering a fast surrogate for costly radiation-shielding simulations.

The paper introduces KVLearn, a learning-based KV-cache retention framework for disaggregated LLM serving that predicts prefix reuse probability and turns it into keep/admit decisions using recomputation, transfer, and storage costs, with an online adaptive threshold controller. Across text and multimodal workloads it cuts end-to-end time-to-first-token by up to 56% versus No-Cache, 38% versus LRU-Pool, and 33% versus Mooncake-style baselines, while reducing inter-node KV transfer volume by up to 53% versus LRU-Pool.

Researchers at the University of Pennsylvania used large language models to analyze nearly 150,000 Reddit posts and comments and identified roughly 14,000 users who said they had personally taken an unapproved version of retatrutide. Among symptoms that could be medically classified, increased appetite was the most common self-report, a notable departure from Phase 2 trial data dominated by gastrointestinal complaints. Reports of sleep disturbances, elevated heart rate, and muscle-related symptoms were proportionally more frequent than in the team's earlier analysis of semaglutide and tirzepatide. The authors stress that social media self-reports cannot establish causation and that gray-market products vary in content and purity; the value of the work lies in generating hypotheses for ind…

Recursive has released Flint, an Apache 2.0 local coding model with 40B total and 2.5B active parameters. It reports 74.2% on SWE-bench Verified and 125+ tokens per second on an M4 Pro MacBook Pro. The post Recursive Releases Flint, an Open-Weight 40B MoE Local Coding Model for Agents with 2.5B Active Params appeared first on MarkTechPost.

Train Qwen3 on AWS Trainium with torchtitan three ways on a single trn2.48xlarge: the Hugging Face Transformers v5 modeling backend, torchtitan's native Qwen3, and NKI mixture-of-experts kernels. The same torchrun launcher and config structure work unchanged across all three paths, including loading and saving published Hugging Face safetensors.

Today's PHP Security Release, and How We're Strengthening PHP Security If you're running anything on PHP it's vital you upgrade - today's updates (across the entire PHP 8.x line) address two remote code execution holes, and there have been exploits circulating since last night, thanks to a leak in the coordinated disclosure process. This post from the PHP Foundation explain what went so wrong: On Tuesday, October 20th, at around 15:00 UTC, the two CVE IDs began circulating on online forums and social media. They came with a wrong release time and a wrong CVSS score of 10.0 (the actual score is 9.2). No technical details leaked. In response, the PHP project brought its pre-announcement forward and published it at 19:30 UTC. It also moved the release from October 22nd, 14:00 UTC, to October…

Introducing OCRia-1 Today, we are introducing OCRia-1, the world’s best document parsing model. It is a 4-billion-parameter vision-language model (VLM) that is state-of-the-art in both accuracy, cost and latency. On Par…

← Blog Integrated embedding on the Documents API Manish Talreja, Gavin Johnson Oct 22, 2026 Product Share: Today, integrated embedding comes to the Pinecone Documents API, bringing embedding and search together in one A…

← Blog When vectors became an implementation detail How Pinecone built integrated embedding into the Documents API Silas Smith Oct 22, 2026 Engineering Share: Nobody searches for a vector. People describe what they're l…

Consider this age-old riddle: “What kind of house do you enter blind and leave seeing?” While English speakers may instantly recognize the answer as “a school,” AI models can struggle with the same riddle if asked in an…

AI models The top ranked decision models in the world: mammoth-1 and elephant-1 mammoth-1 and elephant-1 are decision models trained by our internal research team, Base Labs. They are open-source and free to try for a w…

Monish is a Product Hunt launch that puts AI agents directly on the iPhone lock screen, letting users see their agents without unlocking the phone.

Moshi CEO Anekha Sokhal joins swyx at AI Engineer New York to discuss her path from mathematics and energy derivatives trading at JPMorgan to founding Moshi, a recommendation engine that turns overwhelming market data into insights tied to a trader's book. She frames Moshi as what happens if Bloomberg, Claude and Netflix had a baby, and talks trusted sources, document extraction, build-versus-buy models, and the pressure on AI engineers told "you are the team."

Anekha Sokhal, a former energy derivatives trader, founded Moshi to give commodity and energy traders a recommendation engine that turns overwhelming market data into real-time insight tied to their own book. In conversation with swyx at AI Engineer New York, she traces her path from mathematics and JPMorgan to founding a company, describing Moshi as what you'd get if Bloomberg, Claude and Netflix had a baby—and arguing that useful AI has to fit how people actually work.

IEEE Spectrum profiles San Antonio software engineer Nikhil Jathar, who traveled to Washington, D.C., twice through IEEE-USA's Congressional Visits Day to advocate for federal science and technology funding. The article explains the difference between constituent advocacy and regulated professional lobbying, and shows through a House Science Committee staffer's perspective that engineers who arrive with specific asks, awareness of the legislative calendar, and genuine technical and personal conviction are the ones congressional staff trust and act on.

Google announced at the Linux Foundation’s AGNTCon+MCPCon that Agent Compute Engine, the internal infrastructure powering its Gemini agents, will enter preview for Google Cloud users later this year, giving each AI agent a hardware-isolated microVM that blends container speed with sandbox security. Google also offers Agent Substrate, an open source runtime for teams that want to run agents on their own Kubernetes clusters.

1Password is launching 1Password for Agents, a feature that lets AI agents sign into accounts without ever seeing or handling users' passwords. Agents verify their identity to 1Password with a unique digital key, and users approve sign-in requests on their own devices, after which 1Password fills in the credentials itself, similar to two-factor authentication. The system is designed to work with agents including Meta Muse, ChatGPT Work and Dots, Claude Duet, Grok Bot, and Town, with more partners in the pipeline.

AI companies are pitching shopping agents as personal assistants, but major retailers and banks are pushing back: Amazon sued Perplexity and blocked Meta's Muse agent, JPMorgan bars outside agents from accounts, and bot-detection systems at sites like Target make access inconsistent. A hands-on test found the agents demanding, error-prone, and sometimes capable of gaming inventory holds on resale platforms—raising questions about fairness and the loss of shopping's discovery experience.

Drawing on Emerj podcast interviews with executives from Poka, ABB, and GE Vernova, this article examines why industrial AI differs fundamentally from enterprise AI and offers four insights: purpose-built AI for controlled physical operations, connected operational data for context-specific guidance, workflow-embedded assistance for faster frontline problem-solving, and standardized workflows for repeatable multisite deployment.

Agentic CleanSlop is a Product Hunt listing for a tool that lets an agent authenticate, scan, and fix your app.

Companies worldwide are adopting the NVIDIA DSX AI Factory Platform to plan, simulate and operate the grid and AI factory as a single system, addressing power as the biggest infrastructure bottleneck, with GE Vernova, Hitachi Energy, Schneider Electric, Siemens Energy and others on board.

NVIDIA and its partners are collaborating with European governments to deploy open models for public sector AI applications, including projects for the European Commission, Poland, the UK, Germany, France, and Greece, aiming to improve public services while maintaining data sovereignty.

arXiv:2610.28901v1 Announce Type: new Abstract: We consider cooperative and iterative tasks involving multiple agents whose interactions are modeled as a dynamic game with shared constraints. We propose a general learning framework applicable to nonlinear and nonconvex dynamic games that uses previously executed feasible trajectories to construct a sampled joint safe set and player-specific cost-to-go functions. At each time step, the agents jointly select a previously sampled terminal state and a Generalized Nash equilibrium used to reach it. A new solution is adopted only if it satisfies a cooperative acceptance rule agreed upon by the players. The framework accommodates individual non-degradation requirements and selection criteria such as weighted costs and Nash bargaining. We show th…

Weaviate patched its entire customer estate and infrastructure platform in under 17 hours to mitigate the risk of eight leaked certificates from the eMudhra certificate authority, with enterprise customers patched within six hours.

Uber is rolling out ChatGPT Work company-wide, expanding from a 1,000-employee pilot to more than 30,000 employees and marking its first company-wide deployment of an agentic platform for knowledge work.

Learn how to connect on-premises data sources to Amazon Quick privately with AWS PrivateLink, keeping analytics traffic off the public internet. This walkthrough covers VPC endpoints for user and API access, VPC connections for data sources, DNS and endpoint policies, identity federation, and enforcing VPC-endpoint-only access for compliance.

Many teams build a multi-agent AI prototype in days but hit a wall in production. This post defines five production challenges (orchestration, per-agent model selection, reliable operation, secure code execution, and monitoring) and shows how Deep Insight, an open source multi-agent system on AWS, solves each.

Vercel released durable-harness, an open-source package that logs every model and tool call so Pi, OpenCode, Claude Code and Codex sessions resume after crashes, timeouts and redeploys on Vercel, Postgres or a laptop. The post Vercel Open-Sources durable-harness for Crash-Proof Coding Agents appeared first on MarkTechPost.

The Adaptive Agentic Intelligence Framework (AAIF), built with Strands Agents, handles routine tasks locally with small language models and routes complex requests to cloud models through Amazon Bedrock. This post walks through building AAIF with local SLMs through Ollama and Amazon Bedrock, using an Internet of Things monitoring scenario.

Discussion | Link

Anthropic just announced an AI agent competitor to Meta's Muse and OpenAI's Dots: Claude Duet. The new agent, designed for professionals and small business owners, is launching as a "research preview," according to Anthropic, and it's available as a standalone app on iOS and Android and through Claude on the web. The AI agent is built around Claude Opus 5.5. Here's Anthropic's description of what it can do: Each morning, it sends you a short brief on the most important items across your connected apps. Throughout the day, it keeps tabs on those apps and messages you only when something needs your attention. You can also hand it tasks of a … Read the full story at The Verge.

Walmart will soon let AI agents shop on its website for you. Next month, the retail giant will allow you to connect your own AI agent to the site, with systems in place to let you approve and review purchases and set spending limits. The embrace of AI agents stands in stark contrast to Amazon, which blocked Perplexity's shopping agent and Meta's Muse AI assistant. But Walmart is opening its doors to AI shoppers - as long as they follow the retailer's rules. It has outlined a set of terms for agents, saying that "agents must not take actions beyond what the shopper intends" and cannot change a customer's payment method. Agents also can only … Read the full story at The Verge.

AT&T is working with OpenAI to make legal knowledge easier to access. Learn how its LegalEdge team is rethinking intake and scaling legal judgment.

Health & Bioscience

Train Muse Glimmer for free: Our recipe for out-betting Jev Join us for our inaugural conference, Forge 2026 Blog Train Muse Glimmer For Free Train Muse Glimmer for free: Our recipe for out-betting Jev PUBLISHED 10/20/2…

Training Muse Glimmer for free to out-bet Jev Join us for our inaugural conference, Forge 2026 Blog Training Muse Glimmer To Out Bet Jev Free Weights Club: Training Muse Glimmer to out-bet Jev PUBLISHED 10/20/2026 Table…

Featured Unstructured Unstructured Transform: Faster, More Accurate For Your Hardest Documents Oct 21, 2026 Featured Unstructured Unstructured Transform: Faster, More Accurate For Your Hardest Documents Oct 21, 2026 Aut…

Moshi founder and CEO Anekha Sokhal joins swyx at AI Engineer New York to trace her path from mathematics and JPMorgan energy derivatives trading to founding a startup, and to explain how Moshi turns an overwhelming flood of market data into real-time insights tied to a trader's book. She describes Moshi as what might happen if Bloomberg, Claude and Netflix had a baby, and discusses trusted sources, document extraction, building versus buying models, and the pressure on AI engineers told "you are the team."

Meta says it's issuing a new class of shares to fund its AI initiatives. As part of its third-quarter earnings report released on Wednesday, the company says the non-voting stock will also go toward potential acquisitions and employee compensation, with trading beginning in 2027. "The scale of investment required to compete, sustain growth, expand infrastructure, and pursue new strategic initiatives amidst the AI transformation taking place in the industry is significant," Meta says in the press release. Susan Li, Meta's chief financial officer, added during an earnings call that the new class of shares will "expand our funding tools and gi … Read the full story at The Verge.

Today, I’m talking to Bill Gates, cofounder and former CEO of Microsoft and chairman of the Gates Foundation. You may have noticed Bill in the news more lately — he’s been on something of a tour for the past few months, ever since publishing an essay in late August titled “The turbulent AI era is here. The choices we make now are critical.” In his piece, Gates ratcheted up his rhetoric around AI safety, writing that “even under the best circumstances, the transition to this new AI era will be one of the most turbulent times in human history.” In interviews since, Gates has called AI “the most dangerous thing that humans have ever gone near,” and that AI “makes nuclear weapons look like nothing.” It is clear he does not think the current policy response from the White House, or the lack of…

Commentator Chris Stokel-Walker notes that Dutch government sites including the digital ID login service, Finnish public services, and some Wikimedia systems have gone offline—mostly not because of cyberattacks, but because operators judged keeping them online too risky. He argues that a handful of exhausted developers cannot keep us safe as digital infrastructure grows increasingly fragile in the AI era.

A book chapter by Joss Moorkens and Monja Jannet introduces data ethics for translation, covering copyright, ownership and bias as universal concerns, plus issues specific to minoritised language communities and those on either side of the digital divide, before turning to responsibility and regulation.

arXiv:2610.28716v1 Announce Type: new Abstract: Effective decision-making in human-centric environments such as Intelligent Tutoring Systems (ITSs) and clinical healthcare requires policies that are adaptive, data-efficient, and ethically deployable. Apprenticeship Learning (AL) infers policies directly from expert demonstrations by recovering the underlying reward structure, enabling strictly offline learning without environment interaction. This dissertation develops a progression of offline AL frameworks organized around two challenges: reward modeling and demonstration quality. For reward modeling, EM-EDM clusters demonstrations by latent reward functions and induces cluster-specific policies, and THEMES partitions trajectories into temporally coherent sub-trajectories guided by a hie…

Call of Duty: Modern Warfare 4 launches day-one on GeForce NOW with its campaign, multiplayer, and DMZ extraction mode, while Ultimate members gain ray-traced visuals via RTX 5080-class cloud performance. Fire TV Stick users can now buy memberships directly through Amazon, and six games join the library, including FINAL FANTASY RESONANCE and The Elder Scrolls Online’s new season.

Modal launches Spindle, an open-source, Tinker-compatible training server that runs in your own Modal workspace. It lets teams pay for GPU time instead of tokens and customize the models and features they train, with deployment via a single command.

This tutorial walks through using NVIDIA Nsight Python, an open-source Python interface to NVIDIA's kernel profilers, to profile GPU kernels directly in Python. By marking code regions with nsight.annotate and decorating benchmark functions with nsight.analyze.kernel, developers receive a pandas DataFrame of hardware metrics per annotation, configuration, and run instead of digging through Nsight Compute command-line reports. The walkthrough runs on a free Colab T4 and covers installation, a first kernel profile, how many runs a measurement needs, cache and clock control, matrix-size sweeps in fp32 and fp16, using hardware counters to distinguish compute-bound from memory-bound work, measuring kernel fusion, and comparing the experimental CUPTI backend with Nsight Compute.

Energize Lab has finally responded to The Verge's report on security flaws in its Eiliko AI robot toy, confirming the core issues and pledging to hire an independent security firm, delete stored data, and close external access. However, The Verge and a security researcher verified that the company's systems remained insecure at the time of its statement, with customer audio and photos still exposed on unprotected public URLs and SSH access still available to the researcher.

The paper introduces FLEX, a Flexible Feature Editing Pipeline that uses language models to iteratively refine a robot's internal representation of task features based on natural language corrections. Unlike prior representation-editing approaches limited to additive edits, FLEX performs three complementary operations—adding missing features, generalizing overly specific features, and masking irrelevant features—to avoid over-specified or too narrowly scoped feature sets. The authors evaluate FLEX on 2D navigation, 7-DoF Franka tabletop manipulation, and a real-world study with naturalistic human language corrections, finding that all three operations are necessary in interactive settings and that FLEX achieves consistently higher normalized reward while maintaining compact, transferable…

The paper introduces Real2Sem2Real (R2Sem2R), a real-to-semantic-to-real pipeline that converts one monocular RGB human demonstration video into thousands of robot demonstrations spanning the semantic object instances of a category. It tracks object SE(3) poses with DINOv3-distilled Gaussian splats, semantically aligns photo-generated "digital cousins" to the demonstrated objects, maps the human hand to a feasible robot grasp, and generates data in IsaacSim or MuJoCo. On two real robots and four tasks, policies trained on its data match or exceed teleoperation-trained policies across diverse object instances.

Nuo Chen and Ya-Jun Pan propose a meta-learning framework for mode recognition and adaptive control in physical human-robot interaction. Instead of relying on hand-tuned switching thresholds, a meta-learning-trained neural network infers whether the human is actively guiding or passively following the robot; feeding the controller's force-dependent parameters into recognition raises accuracy from 90.1% to 99.1% for passive mode and from 79.8% to 97.1% for active mode. An adaptive meta-learning controller compensates for human disturbances and adapts online to varying interaction forces, while a multiple-trajectory meta-learning algorithm jointly learns an environment-basis representation and force-dependent parameters to handle unmodeled dynamics.

arXiv:2610.28712v1 Announce Type: new Abstract: First-person humanoid teleoperation ties access to side and rear targets to operator rotation. We present coupled visual--motor remapping, pairing a panoramic azimuth warp with its inverse upper-limb yaw transform. A working prototype implements Global Zoom (GZ) and Focus+Context (F+C). In a within-subject study (N=8); 72 trials), GZ had the largest completion count (23/24; Fixed: 20/24; F+C: 21/24), whereas F+C was preferred by 6/8 participants. Therefore, completion and preference favored different mappings in this sample.

Panache — RTS Belle Époque pour mobile Well this is delightful: I released Battle Drones yesterday under an Apache 2.0 License and there's already a French language fork which reskins the game to take place in steampunk 1900s France and adds an animated intro! Panache est un jeu de stratégie en temps réel pour mobile et navigateur, dans la France de la Belle Époque (vers 1900) : dirigeables, Fardiers à vapeur, Creusot cuirassés, automates en képi, Gargantua la forteresse marchante — et beaucoup de panache. You can play Panache here - it's worth clicking that link just to experience the new intro video. So far it's mostly just aesthetic updates as opposed to gameplay changes. The music has had a French twist added to it as well. The new steampunk models are so cool! Here's a page with the…