Industrial AI should not be confused with traditional process control, according to a Forbes Councils column by Gregory Shahnovsky. The piece distinguishes the two approaches but the source provides n…
Energy companies are building AI Centers of Excellence to move enterprise AI experiments out of pilot purgatory, according to a Forbes Councils column by Abhay Gupta. The column states that no industr…
Rajesh Gharpure argues in a Forbes Innovation CommunityVoice piece that organizations must assess readiness across three areas before increasing autonomy in agentic AI systems, identifying governance …
McKinsey & Company tracks AI token consumption on a user-by-user basis and emails consultants when their usage gets too high, Debasish Patnaik, who leads the firm's UK QuantumBlack AI, data, and analy…
Researchers introduced AutoLexSteer, described as the first fully automated process for building steering vectors for large language models, alongside a new steering vector type called Hangman that op…
Researchers proposed Diverse Skill Routing (DSR), a diversity-aware reranking framework that uses a Determinantal Point Process to balance relevance and non-redundancy when routing user requests over …
A federated compendium of typed, schema-validated artifacts routes within 1.1 points of a centralized system at 20 MB of JSON per client per round on StableToolBench's 3,180 tools, according to an arX…
Researchers proposed DrugReason, a multi-view reasoning framework that integrates knowledge graph-based reasoning with LLM-generated mechanistic inference for drug repurposing, according to an arXiv p…
Researchers introduced PCSDiff, a cascaded task-decoupled diffusion framework for 10-day precipitation bias correction and downscaling, according to an arXiv paper (arXiv:2609.06942v1). Evaluated agai…
A new benchmark called MERIT (Memory Evaluation for Realistic Instrumented Tasks) shows that long-term memory lifts dependent-task success for tool-using LLM agents from a leak-verified floor of 0.00 …
A study of seven experiments on live shared LLM-serving systems found that KV-cache timing side-channel reliability collapses under multi-tenant contention, with mean Cohen's d on a vLLM server runnin…
A projection-based, black-box method operating directly on output embeddings recovers meaningful chronological and spatial structure from language-model representations, according to arXiv paper 2609.…
Researchers introduced a noisy-space action-value (Q-)function and a noisy-space policy gradient (NSPG) that trains diffusion policies for offline reinforcement learning without backpropagating throug…
EdgeMem, a new agent-memory method that preserves original interaction turns in a multi-anchor hypergraph built by lightweight local processing, achieved the highest strict-judge score among seven rep…
A paper on arXiv proposes THGT-FD, a Temporal Heterogeneous Graph Transformer for credit card fraud detection that represents each transaction with one transaction token and six types of relation toke…
A study evaluating fourteen LLMs found that tool-using agents overtrust unreliable tool returns, with mean adoption of corrupted content exceeding one third for every tool and reaching 68.0% for web s…
A black-box attack called AgentLeak can clone a stronger proprietary LLM agent's task-solving capabilities onto a weaker attacker-controlled agent, improving task pass rates by over 40% versus direct …
Researchers posted arXiv:2609.07137v1, a paper introducing Flow3D-OPD, a two-stage post-training framework that applies multi-teacher on-policy distillation to image-to-3D generation built on flow-mat…
A new arXiv paper, arXiv:2609.07153v1, introduces FreqBLiMP, a frequency-controlled extension of the BLiMP benchmark that regenerates all 67 paradigms under explicit Zipf-frequency regimes while prese…
A new arXiv paper (2609.07160v1) formalizes In-place Instruction Following (IIF) for diffusion large language models and introduces GRAFT, a post-training framework that raised the average IIF score f…