The Archive

A pop-up shop for "dots," a personal assistant agent at OpenAI DevDay 2026. (Photo by Heather Diehl/Getty Images) | Getty Images At OpenAI's annual DevDay conference, the company pulled out all the stops to compete with its rivals - primarily Meta, whose Muse AI agent platform has seen early runaway success. CEO Sam Altman walked onstage to cheers and announced Dots, a "real-deal AI" agent powered by GPT-6 Astra, "inspired by the cool agents that we all watched in movies growing up." Similar to Muse, Dots aims for disarming cuteness, taking the form of colorful, personalizable blobs with eyes...

The startup helps developers build AI agents that work over iMessage, SMS/RCS, email, and other messaging platforms. It's a bet that consumers will increasingly use agents instead of downloading apps.

Today I’m talking with Josh Dzieza, a longtime features writer here at The Verge, about Kevin O’Leary’s plans to build a massive data center in Utah. The idea was to build the world’s biggest data center — a 40,000-acre AI campus with nine gigawatts of power, or more than double the average power usage of the entire state of Utah. The project is technically called Stratos, but it’s more prominently known as Wonder Valley, a reference to O’Leary’s nickname on Shark Tank. Josh has spent months reporting on this project, and it’s fair to say Wonder Valley has completely upended Utah politics. Wh...

The glasses stem from the startup’s goal of making hearing care more accessible by addressing the cost, comfort, and stigma associated with traditional hearing aids.

Satlyt wants to be the Android of orbital computing, offering open software that works on many companies' satellites, versus SpaceX's closed, all-in-one iPhone-style approach.

Stratechery interview with Jason Del Rey on Amazon vs. Meta retail competition; not AI-focused.

Gacha Decoding: inference-time method for diverse LLM outputs achieving 2.4x diversity improvement over baselines across chat, creative writing, planning, protein design.

SupraTITO: transferable generative molecular dynamics framework for peptide self-assembly prediction, accelerating exploration of sequence space.

Generation provenance framework for synthetic speech datasets binding source specification, content, waveform, and quality signals for auditable ML training.

Minimax optimal regret analysis for causal logistic bandits under counterfactual fairness constraints with unknown parameters.

Action-on-item schema for unified personalization across movie, news, dialogue histories using shared state-update mechanism.

Self-supervised learning for commute-time-preserving world models enabling latent-space planning via graph Laplacian embeddings.

Hierarchical reduction of piecewise-linear RNNs for interpretable nonlinear dynamical system identification from time series.

LLM-based multi-agent control for adaptive manufacturing: offline sequence generation and online fault handling via OPC UA integration.

Joint nonnegative matrix trifactorization for community detection in multilayer networks across temporal/contextual layers.

Fold'EM: direct atomic structure inference from cryo-EM particles, bypassing density reconstruction to reduce sample complexity.

Port-Hamiltonian neural networks extended with Bregman divergences to model systems with multiple stable equilibria.

Discrete Wasserstein geometry framework for one-step generative modeling on finite state spaces with amortized training.

LLM-generated scientific text shows different argumentative structure than human writing under Swales' CARS model.

PACE system enforces safety for tool-using LLM agents by tracking provenance and mediating tool calls before execution.

Claim-specific verification audit for modular agents replaces aggregate scoring with evidence-based component evaluation.

ARCCS automates regulatory compliance checking by decomposing legal obligations into traceable requirements with evidence grounding.

Reachability-informed reinforcement learning for spacecraft trajectory design with learned waypoint selection.

Robust scheduling framework uses classification models to improve non-clairvoyant job completion time minimization.

European Crisis Management Ontology (ECMO) defines modular OWL-based schema for cross-sectoral disaster management.

PPO-HRAP combines proximal policy optimization with regime-aware policy for risk-controlled trading with drawdown penalties.

Biomedical reranking improves evidence selection in RAG pipelines for clinical QA over longitudinal patient records using PubMedBERT and MedCPT.

TRACE bounds multi-turn jailbreak risk in LLMs via trajectory analysis, showing single-turn safety training alone cannot prevent multi-turn exploitation.

Clifford Sheaf Neural Networks apply Clifford algebras to equivariant graph learning with K-term sandwich restriction maps.