Who answers when an agent acts?
Banks are triaging AI voice agents, a legal site added AI arbitration to its terms, and ModelOp now enforces policy at runtime and at the kernel level.
30 stories, 180 Manifests cited.
Banks are triaging AI voice agents, a legal site added AI arbitration to its terms, and ModelOp now enforces policy at runtime and at the kernel level.
New research turns video diffusion models into action learners, builds a generalist simulator from 30,110 hours of robot data, and warns that prediction error alone does not prove a model understands physics.
Four new frameworks let agent teams evolve their organization, let foundation models improve without a teacher, and let vision-language models practice on their own generated data.
New evaluation work finds that zero-shot robot rankings predict fine-tuned success poorly, that popular benchmark scores are unreliable for comparing models, and that partial progress tells more than pass or fail.
Several new studies and one new model converge on the same point: running every request at maximum reasoning effort wastes tokens, and lower effort often costs much less with little lost.
ByteDance released an open agent foundation model with its full recipe, Sarvam released a fast decision model with open weights, and Snowflake made fine-tuning open-weight models serverless.
Larridin raised $30 million to measure AI value, Enderis argues training data is the real enterprise bottleneck, and Salesforce acquired Cresta to scale AI customer service.
Attackers can learn what trained a model, copy its capabilities, and turn routine operational data into instructions, all by querying the systems companies already expose.
Supermicro will build systems pairing SambaNova chips with NVIDIA GPUs, Accelsius is supplying direct-to-chip cooling for energy research, and Astera Labs reported record connectivity revenue.
A16z's Martin Casado argues AI is an industrial-scale shift, Fast Money's panel sees fatigue in the AI trade, and developers are moving to multi-model, increasingly self-hosted portfolios.
SpecPilot discovers decoding optimizations automatically, MoE pruning research shows replacement choice matters, and new compression and distillation methods squeeze more from smaller models.
WebHarbor offers 153 dockerized websites for stable agent training, a 4B hybrid agent mixes GUI with terminal and code, and LangChain cut browser-agent cost by rebuilding its browser stack.
Switching between unfinished tasks costs coding agents 11.5 points of performance, a new approval system checks commands by their effects, and harness tuning now trades quality against cost explicitly.
In preprints on blinded AI peer review, reasoning failures in bioinformatics sandboxes, traceable proteomics and statistically controlled neural analysis, the pattern is the same: trust has to be engineered, not assumed.
In preprints on solubility, blood-brain barrier permeability, ADMET endpoints and metabolite identification, the same lesson recurs: how a molecule is described shapes what a model can learn from it.
ING's 2027 rates outlook sees US yields rising toward 5.25 to 5.5 percent before easing, euro rates volatile with a constructive year-end, and sterling drifting lower. New Zealand's soft labour data and Türkiye's sticky inflation show why the path is not smooth.
Wayfair reported $3.4 billion in revenue, LGI Homes delivered 10.4 percent more homes, and Marriott Vacations raised its guidance after a hurricane. Smith Douglas Homes, meanwhile, earned just $0.07 a share as margins compressed. The consumer is not one story.
From home goods to diagnostics to travel, third-quarter reports kept landing with a raised full-year outlook attached. The quarter itself was mixed. The message about the rest of the year was not.
Three quarters of surveyed Italian banking leaders say their institutions lose more than $10 million a year to fraud. In Brazil, fake-lawyer scams are driving a surge in social engineering. And the Binance theft keeps moving through laundering chains months later.
Concentrix, Pfizer, Erie and BGC all disclosed finance or chief-executive leadership changes in the same batch of filings, and the pay disclosures next to them show boards paying up for continuity. One departure named no successor at all.
China's central bank eased support for share buybacks while tightening foreign-exchange liquidity, India's central bank ran two overnight liquidity auctions in a day, and Spanish firms still face rising loan rates. The action is in the plumbing.
In preprints on a nerve-stimulation reflex for brain injury, acoustic stimulation for sleep rhythms in older adults, electrical stimulation after adolescent alcohol exposure, and a Parkinson's sleep trial design, the tool is stimulation, timed and targeted.
In preprints on cardiac enhancers, skin glycogenes and a fat-to-skin wound signal, age-related regulatory changes look less like fixed damage and more like states that can shift.
The Jets erased a four-goal deficit, the Rangers ended a four-game skid and the Ducks extended a scoring run into franchise history.
From waived cornerbacks to a reported quarterback signing, clubs reshuffled rosters ahead of Week 9, balancing injuries against practice-squad depth.
Seattle and Houston changed leadership, Boston and St. Louis swapped pitchers, and several clubs laid out what they want from the free-agent market.
Two bills became law, two proclamations set out positions on space and ideology, and the administration promoted manufacturing growth and a busy October travel schedule, with a diplomatic meeting set for November 5.
First and Second Circuit rulings, a Ninth Circuit stay decision, a consent decree, an extradition and two ICE enforcement encounters make up the federal justice docket.
In preprints on flood-resilient wastewater planning, a 70-year reservoir sediment record, sediment modelling and field water-quality testing, the shared lesson is to measure the system you have, not the one you assume.
In preprints on a battery electrolyte roadmap, a free-energy mapping toolkit, proton conductors under pressure and ion-selective membranes, the shared question is where ions can go, and how fast.