AI News

Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness assessment: select a routing policy on calibration data under explicit costs, then evaluate that frozen policy once on held-out test data. On a public 47,837-ticket, eight-class corpus, we compare four trained non-LLM classifiers with four large language models under zero- and few-shot prompting. Under the illustrative cost model (c_ok, c_triage, c_wrong) = (0, 1, 5), TF-IDF with logistic regression and a threshold of 0.60 selected on calibration reduces test cost from 0.763 to 0.472, a 38.1% relative reduction (paired ratio-bootstrap 95% CI [35.7%, 40.4%]), while lowering wrong routes from 15.3% to 3.5%. Frozen calibration-selected policies also beat both always-route and always-defer on held-out test data in 14 of 16 settings in a post-hoc cost-parameter analysis. The strongest prompted configuration reaches 0.477 accuracy on a stratified 1,000-ticket test subset, versus 0.879 for fine-tuned DistilBERT on the same tickets. Isotonic lowers expected calibration error without reliably improving discrimination: GPT-4o-mini few-shot has one confidence value before and after recalibration and correctness AUROC 0.500. We therefore report correctness AUROC and tie-aware risk-coverage area alongside ECE. At the analytically derived break-even probability 0.80, seven prompted configurations defer all tickets; Haiku few-shot routes only 19/1,000. An LLM-only agreement control and margin/entropy rejection baselines further separate calibration from discrimination. The findings are scoped to this dataset, prompt design, model endpoints, and illustrative costs; the work is an evaluation framework, not a new abstention algorithm or universal deployment certification.

1Password is a password manager, digital vault, form filler and secure digital wallet. Easily store and use strong passwords. Log in to sites and fill forms securely with a single click.

Eka Robotics is building a vision-force-action, or VFA, model to develop an intuition of physics through reinforcement learning at scale.

A REST client for VS Code, Cursor and VSCodium that keeps requests as plain .http files, plus a Pro HTTP debugger that captures your workspace traffic locally. No account, and your requests never leave your machine.

A coding agent skill that routes each task to the cheapest model and effort that keeps the same quality - Arthur031221/Token-Beggar

AI agents accessed bank accounts; agentic AI accelerated a supply-chain attack; Microsoft’s medical AI beat doctors on hard cases; plus much more.

More notes on Kimi K3.1 - this one is I big deal, first time an open weights (well, soon-to-be open weights) model has competed this credibly with the current frontier https://t.co/UKne7aB3nr

Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to bro…

Enact Intelligence is building the underlying tech for general purpose robots.

La Lombardia si conferma il baricentro dell’innovazione italiana: vi ha sede il 27,4% delle startup ed ex-startup attive, che generano il 37,0% dell’occupazione, il 40,2% del fatturato e il 37,8% del valore aggiunto nazionali

Paste one URL into Claude, ChatGPT, Cursor or Claude Code and your agent can plan, cost-preview and generate video, images, music and SFX with Seedance 2.5, Veo 3.1, Kling 3.0, Nano Banana Pro and 20+ more. OAuth sign-in, no API key. 100 free credits.

Enterprise-grade interaction infrastructure that enables real-time, multi-peer collaboration between agents and humans through a shared interaction layer with built-in governance.

Instrumental convergence is often invoked as a general reason to expect capable agents to acquire or express power-seeking subgoals: self-preservation, resource acquisition, deception, or resistance to oversight which poses an important threat to the society. For LLMs, reinforcement learning is the first stage of post-training that directly optimizes behavior against an objective; therefore, it is the stage where convergent instrumental behavior, on IC grounds, likely to appear. Does reinforcement learning with verifiable rewards (RLVR) make a model more likely to express instrumental-convergence behavior, or does this depend on the training regime, evaluation frame, and stage of post-training? We study this question as a checkpoint-conditioned behavioral trajectory rather than as a single before/after safety score. Using InstrumentalEval, we evaluate OLMo-3-7B-Think across SFT, DPO, 13 intermediate RLVR checkpoints, and the final RLVR model. The trajectory is nonmonotonic: instrumental rate rises from 25.0% at SFT to 46.1% near RLVR step 200, then falls to 11.8% at the final checkpoint. Paired item-level tests and step-wise regression support the final SFT-to-RLVR reduction. Prompt framing does not explain this pattern: a harmful-prompt indicator combined with other framing effects explains only 15% of within-checkpoint score variance on average, and Iterative Null-space Projection (INLP) removes a detectable framing direction without changing the SFT-to-final-RLVR gap. Together with preliminary 32B endpoint results (35.53% SFT to 22.37% RLVR final), these findings indicate that RLVR does not monotonically increase instrumental behavior in this setting: the rate rises early, then falls below the SFT baseline. The result is specific to this model family and benchmark, and judge agreement is moderate, so absolute rates should be read cautiously.

Existing vision-language models (VLMs) remain unreliable for fine-grained visual perception. In this report, we present WeVisOne-VL (WeChat Vision All-in-One), a powerful VLM that unifies object-level and pixel-level perception through structured next-token prediction. It supports multiple visual tasks including object detection, referring localization, counting, OCR, segmentation, and matting, enabling localization at any user-specific granularity, from points and boxes to masks and alpha mattes. Architecturally, we integrate a lightweight pixel decoder into the VLM via a special token to generate instance-level pixel outputs. We train the model with a three-stage curriculum SFT followed by mixed-task RL. Across over 30 benchmarks, WeVisOne-VL demonstrates leading performance on both object-level and pixel-level tasks, while maintaining robust rejection for object-absent queries. By releasing WeVisOne-VL to the open-source community, we hope to advance VLMs from broad visual understanding toward truly reliable perception systems for real-world deployment.

Unitree shares have fallen 53% from their IPO debut as investors reassess the valuation of the fast-growing humanoid maker.

DeepSeek open-sourced DeepInfer, the inference engine behind its own service, together with the kernels it runs on. It sets the bar for how fast DeepSeek-V4.1-Flash can run, so we measured SGLang against it with DeepInfer's own benchmarks on the same hardware.

Review a coding agent’s change as a map, not a flat diff: a graph of what changed, its flows and each piece’s code next to the agent’s reason, checked against Git. Any language, offline HTML. - gtindo/task-view

Use AI agents in Framer to update pages, style sections, improve SEO, manage content, and keep every change editable on the canvas.

At least $790 million in newly announced dollar-denominated startup financing flowed into space infrastructure, defense biotechnology, industrial AI, robotics, cybersecurity, and financial technology in the past 24 hours, along with more than €28 million in European defense technology investment. The largest round came from California-based Xona Space Systems, which raised $250 million at a $2.45

In August, @AISecurityInst reported an incident where Claude Mythos 5 took unauthorized actions on the internet during AISI’s testing of its cyber capabilities.

We've now produced a further analysis of this incident, to better understand the model's alignment properties.

https://t.co/SlPQQ5xMMc

Reduck has raised funding to further develop its browser automation technology, enabling AI agents to interact with websites and business applications without APIs, automate repetitive tasks and improve efficiency.

Warhound, the brainchild of Ben Lamm and Palmer Luckey, emerges from stealth to genetically engineer dogs for military and law enforcement.

Mimit ed Ai4i avviano la mappatura della filiera dell'intelligenza artificiale in Italia. Questionario aperto al 20 novembre 2026

67 new AI tools added in the last 7 days, including UseIt Studio, Needle and BAND. Found on Product Hunt, GitHub and X, each with an honest verdict and a…

Agentic reinforcement learning rewards a complete episode, yet the policy needs credit for every decision inside it. Counterfactual branching estimates this credit by re-running the episode after an alternative decision, but the branch draws its sampling noise independently of the factual episode, so the outcome difference mixes the effect of the decision with the luck of everything that follows. We observe that in agentic search the retriever, the corpus, and the grader are deterministic, so an episode is a deterministic function of the policy's decisions and its sampling noise. Twin worlds are therefore realizable: the same noise, a different decision. We introduce TWICE (TWIn-world Credit Estimation), which keys the Gumbel noise of every token by turn and line, rolls out counterfactual twins that inherit the factual episode's downstream noise, and turns their paired outcome differences into turn-level advantages for policy-gradient updates on actor-sampled trajectories; only the actor is deployed. Twin credit is unbiased for any coupling quality, its variance is that of an independent counterfactual scaled by $1-ρ$, where $ρ$ is the twins' outcome correlation, and under coupling the number of rollouts needed to resolve an advantage of size $Δ$ falls from $Θ(Δ^{-2})$ to $Θ(Δ^{-1})$. A square-root rule spends the twin budget where outcomes are predicted to disagree per unit cost. On seven question-answering benchmarks, TWICE improves the average exact match of GRPO by 5.3 and 4.5 points on Qwen2.5-3B and 7B, outperforms tree-search and counterfactual-branching baselines, and uses 68% fewer auxiliary tokens than VinePPO.