Makoto is an integrity hook for Claude Code that monitors the AI agent's tool calls and blocks those that fake checks, such as claiming to run tests without actually running them, fabricating citations, or disabling security verifiers. It operates on a ledger of the agent's own claims and ensures promises are fulfilled.

Three years ago, Sequoia partner David Cahn was one of the first to quantify the financial implications of Silicon Valley's massive AI infrastructure spending. Starting from Nvidia's $50B GPU revenue, he calculated that $200B in revenue would be needed to pay back the upfront investment.

OpenAI released three new GPT-5.6 models—Sol, Terra, Luna—alongside major app updates, including ChatGPT Work and Codex integration. The models show strong performance on benchmarks at lower costs, with Sol being the most capable. Independent evals confirm near-frontier results, especially in coding and agentic tasks.

Meta's new Muse Image model lets anyone generate AI images of you using your public Instagram handle without notification. This article explains the risks, how to opt out, and additional security measures like enabling MFA and switching to a private account.

In early 2026, multiple AI subscription services reduced quotas and raised prices, causing user dissatisfaction. The article reviews the brutal competition in 2025 and highlights the current trend of service contraction.

A new trend sees AI-focused investors acquiring accounting firms and mandating the use of OpenAI's technology, raising questions about industry disruption and data privacy.

SK hynix, a supplier of advanced memory chips, has seen profits skyrocket thanks to the global race to build AI datacentres. The South Korean chip maker set pricing for its mega US listing on Friday, aiming to raise $26.5bn.

Chancellor Rachel Reeves is to announce a new City “skills compact” that will commit firms such as Barclays and Lloyds to retraining thousands of financial sector workers for the AI revolution. The government-backed initiative will be launched on Tuesday and aims to help workers keep pace with technological changes that have sparked fears of mass redundancies.

AgentKey is a one-stop live data marketplace for your AI agent.

Robbyant, Ant Group's embodied-intelligence unit, has released LingBot-World-Infinity (LingBot-World 2.0), a 14B causal video generation model that acts as an interactive world simulator. Its core innovations—Mixture of Bidirectional and Autoregressive (MoBA) attention and distribution matching distillation—tackle long-horizon drift. A Director-Pilot agentic harness enables infinite video generation. The paper demonstrates a 60-minute session, but the open-source release includes only one checkpoint and a 480P script, lacking deployment code and quantitative benchmarks, under a non-commercial license.

OpenAI launches GPT-5.6 with three models: Sol (flagship), Terra (workhorse), and Luna (fast). Free for all users. Covers pricing, benchmarks, safety, and hands-on tests.

We built a prototype that verifies code migration with a proof instead of tests. A deterministic translator converts both the original and rewrite into Lean, and a theorem prover certifies they produce identical results on all inputs. The approach is demonstrated with the iCPPI portfolio insurance algorithm.

Headlong's adaptation of Karel Čapek's 1920 play RUR brings timely debates about AI and robot consciousness, though philosophical discussions sometimes drag. Co-produced with Schwarzman Centre, Oxford, the play is informed by cutting-edge Oxford research.

Multi-robot localization requires accuracy and consistency. Centralized approaches are optimal but impractical. This paper proposes D-CLIPSE, a distributed consensus-based filtering framework that shares preintegrated odometry and shared states, achieving near-centralized performance with better consistency.

This study systematically evaluates the feasibility of current humanoid robots for laparoscopic surgical tasks. The researchers developed a humanoid-based teleoperation framework using general-purpose instruments, and assessed its performance through benchtop characterization, dry-lab user studies with surgeons of varying experience, and in vivo porcine studies. The results highlight both the promise and key technical challenges that must be addressed before clinical deployment.

Researchers present a personalized pneumatically-actuated soft robotic exoglove using topological scans for custom fit, FEA for pHRI force analysis, and pneumatic control for precise joint mobilization, laying groundwork for dexterous rehabilitation.

Researchers have developed a novel soft robotic exoglove designed to simultaneously improve hand mobility and alleviate muscle spasticity through massage-like compression. The glove uses personalized soft pneumatic actuators tailored to individual hand topology and kinematics, with preliminary experiments validating comfort and effectiveness. This technology could benefit the 12 million people worldwide affected by hand spasticity and associated pain.

This paper proposes a unified monocular vision-based grasping framework that handles both soft and rigid objects using a single control pipeline with only RGB input and a position-controlled gripper. It combines open-vocabulary detection, segmentation, point tracking, and depth estimation, and uses a language-based stiffness estimation model to select the grasping strategy. Experiments on a Franka Emika arm successfully grasped lettuce, mozzarella, croissants, paper towels, and plastic bottles.

A data-efficient and interpretable method for vision-based dynamic obstacle avoidance using pretrained models (UniDepth, SuperPoint, SuperGlue) that computes per-keypoint time-to-collision (TTC) to select evasive motion. Evaluated on M3ED dataset, achieving 0.49 precision and 0.38 recall for detecting TTC<1s frames, and detecting 20 out of 22 obstacles. No model training required—only 74 seconds of data for hyperparameter tuning.

STEMbot is a miniature climbing robot designed for autonomous navigation under plant canopies to enable early pest detection. It integrates PIN-SLAM and a semantic OcTree, and uses a manifold-constrained A* planner, demonstrating reliable traversal on stems of 7-33mm with reconstruction accuracy under 1cm.

Shift & Drift is a dual-track benchmark that evaluates autonomous driving motion planners under semantic distribution shifts (novel urban topologies) and state-distribution drifts (execution perturbations). The study finds that imitation learning methods perform well in-distribution but fail under semantic shifts, while reinforcement learning-based planners exhibit graceful degradation.

arXiv:2607.07830v1 proposes a two-stage physics-guided framework called HumoSlope for robust humanoid locomotion on steep slopes. Stage I uses a slope-adaptive ZMP regularizer for a terrain-consistent balance prior; Stage II introduces a Biomechanical Slope Gait Adapter that dynamically modulates CoM height and limb coordination based on estimated slope geometry, avoiding crouched gaits. Sim-to-Real experiments show blind traversal of outdoor grass slopes up to 62.7% (32.1°).

The paper proposes SASGeo, a semantic map-localization framework using persistent structures like roads and buildings for GNSS-denied UAVs. In 220 retrieval trials, spatial semantic matching variants achieved 94.5-95.5% Recall@1, significantly outperforming global descriptors (58.6%), though variants overlapped. The synthetic proof of concept shows promise but requires real-flight validation.

APIVOT is a VLM-based planner that adaptively interleaves language and visual thoughts for long-horizon robot planning, achieving significant gains in spatially constrained kitchen tasks.

This paper proposes SAGA, a training-free stable acceleration guidance method to improve temporal instability in autoregressive video diffusion. By using acceleration-domain spectral guidance and structured noise initialization, it effectively reduces flickering and jitter, enhancing temporal and image quality.

LightCrafter is a novel hybrid pipeline for video relighting that reformulates the task as video translation of a proxy PBR rendering. It combines the strengths of physically-based rendering and diffusion models to achieve long-form temporal consistency and fine-grained lighting control, outperforming prior state-of-the-art on real-world benchmarks and providing a synthetic benchmark for further analysis.

FedTR combines federated learning and transfer learning to address data scarcity and complexity in industrial visual inspection, achieving high accuracy on label defect identification.

Proposes LOGOS, a novel transformer-based approach that leverages textual prompts to guide oriented object detection in aerial images, outperforming existing methods on the DOTA dataset, especially in dense and rotated scenarios.

The paper introduces ThermoField, a framework that unifies thermal scene reconstruction and thermophysical parameter estimation via differentiable heat-transfer simulation. It uses neural fields to represent spatially varying properties like thermal diffusivity, constrained by scene geometry and physics, enabling joint reconstruction of geometry, estimation of diffusivity, and prediction of thermal evolution under unseen conditions.

Researchers propose adversarial decoys, independently optimized image patches that redirect attention away from adversarial regions, bypassing attention-based defenses in Vision Transformers. The approach decouples misclassification and defense evasion, is attack-agnostic, and preserves attack effectiveness. Experiments on ImageNet reveal a fundamental limitation of using attention magnitude as an indicator of adversarial relevance.