All entries

This post from Replit outlines a 4-step security framework for protecting vibecoded applications in production environments. It emphasizes that application speed must not come at the expense of securi

tutorialsreplit--x

Safety

How VLAs Fail Differently: Black-Box Action Monitoring Reveals Architecture-Specific Failure Signatures

DGX agent

arXiv:2605.28726v1 Announce Type: cross Abstract: We discover that VLA architectures fail in fundamentally different, predictable ways at the motor-command level. Running VQ-BeT, Diffusion Policy, and

safetyarxiv-cs-lg

28 May 2026

How we built Cloudflare's data platform and an AI agent on top of it

Cloudflare describes the architecture and development of their unified data platform that integrates data from across their global network, along with an AI agent built on top to enable intelligent qu

agentscloudflare-ai

HRBench: Benchmarking and Understanding Thinking-Mode Switch Strategies in Hybrid-Reasoning LLMs

arXiv:2605.28398v1 Announce Type: new Abstract: Hybrid-reasoning large language models (LLMs) expose explicit controls over reasoning effort, allowing users or systems to trade off answer quality agai

https://github.com/run-llama/liteparse これか。日本語PDFでどんなか試しとこう。

model-releasesjerry-liu--x

https://github.com/run-llama/liteparse これか。日本語PDFでどんなか試しとこう。 We've created the world's fastest PDF parser ⚡️ And it's more accurate than any other open-source, model-free PDF parser out there (pymupdf

https://mistral.ai/news/ai-now-summit-2026/

model-releasesmistral-ai--x

Mistral AI announced its participation in or perspective on the AI Now Summit 2026, likely discussing developments in AI safety, ethics, or industry trends relevant to the conference. The announcement

Human-AI Collaboration for Estimating Scientific Replicability

arXiv:2605.27394v1 Announce Type: cross Abstract: Determining whether published scientific findings can successfully be replicated is a long-standing challenge in the empirical sciences. Existing appr

Tools

Human Interface Guidelines, Apple.

Apple's Human Interface Guidelines are design principles and standards that establish best practices for creating user interfaces and experiences across Apple's ecosystem of products. These guidelines

toolslinus-lee--x

Tutorials

Human Label Variation as Stable Signal: Learning Annotator-Specific Explanation Behavior via Cross-Annotator Preference Optimization

arXiv:2605.28802v1 Announce Type: new Abstract: Free-text explanations extend human label variation (HLV) beyond label disagreement by revealing the reasoning and preferences behind annotators' decisi

tutorialsarxiv-cs-cl

Human-like in-group bias in instruction-tuned language model agents

arXiv:2605.28114v1 Announce Type: new Abstract: As autonomous AI agents are deployed in persistent, interacting networks -- coordinating tasks, routing resources, and accumulating reputational histori

safetyarxiv-cs-ai

HumanoidMimicGen: Data Generation for Loco-Manipulation via Whole-Body Planning

model-releasesarxiv-cs-ai

arXiv:2605.27724v1 Announce Type: cross Abstract: Imitation learning is a promising approach for training humanoid robots to both walk and manipulate, but it requires a large number of demonstrations,

Hurwitz Quaternion Multiplicative Quantization for KV Cache Compression

model-releasesarxiv-cs-ai

arXiv:2605.27646v1 Announce Type: cross Abstract: We propose extbf{Hurwitz Quaternion Multiplicative Quantization (HQMQ)}, a extbf{calibration-free} method for KV cache compression of large language m

Hybrid Neural World Models

arXiv:2605.28317v1 Announce Type: cross Abstract: Neural surrogates promise large speedups over classical solvers for physical dynamics but fail silently at sharp dynamical events such as shocks, fron

Local Ai

I built a chrome extension which can navigate, scroll, fill forms and type and scrape data on any website

A Chrome extension built for automated web interaction and data extraction that uses Ollama's local LLM capabilities to intelligently navigate websites, scroll pages, fill out forms, type text, and sc

local-air-ollama

“I can’t breathe” were the last words of Henry Nowak as he was handcuffed by British police while heavily bleeding. 🇬🇧💔

I cannot verify this claim with confidence. The post appears to reference an incident involving someone named Henry Nowak and British police, but without access to credible reporting or official sourc

industryelon-musk--x

I had early access to Opus 4.8. Was impressed by it. Here is Opus 4.8's one shot of 'create a visually interesting shader that can run in tw…

model-releasessonya-huang--x

I had early access to Opus 4.8. Was impressed by it. Here is Opus 4.8's one shot of 'create a visually interesting shader that can run in twigl, make it like an infinite city of neo-gothic towers part

I had Opus 4.8 in Claude Code write a sophisticated, if minor, academic paper from a archive of hundreds of de-identified research files fro…

model-releasesethan-mollick--x

I had Opus 4.8 in Claude Code write a sophisticated, if minor, academic paper from a archive of hundreds of de-identified research files from years ago I had to use GPT-5.5 Pro as a reviewer, it spott

Applications

I have spoken to lots of companies that blew through their token budget for the year in months, but a half a billion dollars on internal emp…

applicationsethan-mollick--x

Ethan Mollick discusses companies that rapidly depleted their annual token budgets within months, with some spending hundreds of millions of dollars on internal employee use of large language models.

I Hear, Therefore I Trust: A Socio-Technical Investigation of Humans as Synthetic Speech Detectors

arXiv:2605.28064v1 Announce Type: cross Abstract: Automatic deepfake detection has received considerable research attention, yet the socio-technical environment in which humans actually encounter synt

I interview dozens/hundreds of new grads, nearly every day of the year. These are people with a well-formatted resume and a Bachelor's Degre…

I interview dozens/hundreds of new grads, nearly every day of the year. These are people with a well-formatted resume and a Bachelor's Degree in Engineering from well-regarded US universities and a GP

industryelon-musk--x

I read this garbage (in a big UK newspaper) when Google can’t even count reliably and wonder why people don’t spend more time learning about…

I read this garbage (in a big UK newspaper) when Google can’t even count reliably and wonder why people don’t spend more time learning about AI’s actual strengths and weakness before running their mou

safetygary-marcus--x

I signed up for another SaaS

model-releasesben-s-bites

Ben Kamarov reflects on his decision to sign up for yet another SaaS product, likely discussing his evaluation criteria, the specific tool's features, or lessons learned about SaaS adoption and tool p

I think you’ll really like Opus 4.8 It’s as smart as its benchmarks show but expresses and utilizes that intelligence in a warm and collabor…

I think you’ll really like Opus 4.8 It’s as smart as its benchmarks show but expresses and utilizes that intelligence in a warm and collaborative way. Workflows are a great way to utilize it- I’m hook

model-releasesthariq--x

I tried the liteparse's web browser version today to convert a couple of PDF to text and was shocked at the speed. I had to recheck twice to…

model-releasesjerry-liu--x

I tried the liteparse's web browser version today to convert a couple of PDF to text and was shocked at the speed. I had to recheck twice to see whether it even did the complete processing or not 😅 ht

IAR2: Improving Autoregressive Visual Generation with Semantic-Detail Associated Token Prediction

model-releasesarxiv-cs-cv

arXiv:2510.06928v2 Announce Type: replace Abstract: Autoregressive models have emerged as a powerful paradigm for visual content creation, but often overlook the intrinsic structural properties of vis

IBM and Red Hat commit $5B to establish a new model for open-source software, dubbed Project Lightwell, and will deploy 20,000 engineers, supported by AI (Connor Hart/Wall Street Journal)

Connor Hart / Wall Street Journal: IBM and Red Hat commit $5B to establish a new model for open-source software, dubbed Project Lightwell, and will deploy 20,000 engineers, supported by AI — Project L

model-releasestechmeme

IBM expands digital sovereignty push with new cloud compliance and visibility platform

IBM Corp. today introduced a new cloud sovereignty tool aimed at helping enterprises demonstrate compliance, operational control and governance as artificial intelligence deployments expand across hyb

industrysiliconangle

IBM, Red Hat launch $5B Project Lightwell to boost open-source security

IBM Corp. and its Red Hat subsidiary today launched an initiative called Project Lightwell to improve the security of open-source projects. Project Lightwell is backed by a 5 billion commitment. In ad

industrysiliconangle

ICAN-Deploy: Identity-Stable Canary Deployment for Safety-Critical Embodied Agents

arXiv:2605.28097v1 Announce Type: new Abstract: Canary deployment routes a fraction of traffic to a new software version, monitors metrics, and rolls back on regression. Mainstream controllers (Argo R

safetyarxiv-cs-ro

ICG: Improving Cover Image Generation via MLLM-based Prompting and Personalized Preference Alignment

arXiv:2605.27374v1 Announce Type: new Abstract: Recent advances in multimodal large language models (MLLMs) and diffusion models (DMs) have opened new possibilities for AI-generated content. Yet, pers

safetyarxiv-cs-cl

Tutorials

Identifiable Bayesian Deep Generative Copulas with Unknown Layer Widths for Data with Arbitrary Marginal Distributions

arXiv:2605.27523v1 Announce Type: cross Abstract: Deep generative models offer powerful tools for multivariate data analysis, but their black-box architectures are often unidentified and difficult to

tutorialsarxiv-cs-lg

Identifying and Mitigating Bottlenecks in Role-Playing Agents: A Systematic Study of Disentangling Character Profile Axes

arXiv:2601.04716v3 Announce Type: replace Abstract: While Large Language Model (LLM) role-playing agents have advanced rapidly, it remains unclear which profile elements genuinely drive role-playing q

safetyarxiv-cs-cl

Identifying and Understanding Human Values in Text: A Tailorable LLM-based Architecture

arXiv:2605.27373v1 Announce Type: new Abstract: As intelligent systems become more autonomous, the scientific community focuses on creating decision-making mechanisms that include ethical and moral co

agentsarxiv-cs-ai

Local Ai

Identifying Explicit Parsimonious Piece-wise Polynomial Relationships in Industrial time-series: Application to manipulator robots

arXiv:2605.28320v1 Announce Type: cross Abstract: This paper addresses the problem of identifying parsimonious explicit piece-wise polynomial relationships that might involve a relatively large number

local-aiarxiv-cs-ai

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

researchapple-ml-research

CVPR 2026 is a major international conference on computer vision and pattern recognition organized by IEEE and the Computer Vision Foundation, where Apple presents its latest machine learning research

if you replace billions with millions, this sounds like any other high-growth startup fundraise announcement 😉

model-releasesjerry-liu--x

if you replace billions with millions, this sounds like any other high-growth startup fundraise announcement 😉 We've raised 65 billion in Series H funding at a 965 billion post-money valuation, led by

if you're a VC and want a killer portfolio monitoring tool like ours, but won't want to build it yourself... check out http://cura.inc

if you're a VC and want a killer portfolio monitoring tool like ours, but won't want to build it yourself... check out http://cura.inc this is my sortable table of all portfolio companies with revenue

agentsyohei-nakajima--x

If you're building your own cloud agent like Devin or Ramp Inspect, there's lots of great details here on setting up VMs, computer use, memo…

If you're building your own cloud agent like Devin or Ramp Inspect, there's lots of great details here on setting up VMs, computer use, memory, and more. Fun deep dive with the creator of OpenInspect

agentsswyx--x

IFMTBench: A Comprehensive Benchmark for Multilingual Translation Instruction Following

model-releasesarxiv-cs-cl

arXiv:2605.28218v1 Announce Type: new Abstract: Modern translation workflows demand more than semantic equivalence. Users routinely require models to preserve JSON or HTML schemas, honor curated gloss

Applications

IGADA-IoT: IoT Sensor Energy Optimization in Wireless Sensor Networks Driven by Automatic Data Augmentation

arXiv:2605.27397v1 Announce Type: new Abstract: In wireless sensor networks (WSNs), data augmentation is a novel method to improve sampling-frequency decision performance, thereby enabling energy opti

applicationsarxiv-cs-lg

I’m a big proponent of doing the complete opposite. If you extrapolate current adoption and progress, you quickly realize we’re heading towa…

industrycristobal-valenzuela--x

I’m a big proponent of doing the complete opposite. If you extrapolate current adoption and progress, you quickly realize we’re heading toward a future where almost all video is AI-generated. In that

I'm proud to share that @Glean has surpassed 300M ARR, just five months after crossing 200M and growing ~3x over the past 15 months. This …

model-releasessonya-huang--x

I'm proud to share that @Glean has surpassed 300M ARR, just five months after crossing 200M and growing ~3x over the past 15 months. This is an exciting milestone for Glean, and it's a signal about wh

Imitating and Finetuning Model Predictive Control for Robust and Symmetric Quadrupedal Locomotion

arXiv:2311.02304v3 Announce Type: replace Abstract: Control of legged robots is a challenging problem that has been investigated by different approaches, such as model-based control and learning algor

safetyarxiv-cs-ro

Imitation Learning for Robot Assistance in Open Surgery: A Multi-Policy Evaluation on Suture Following

model-releasesarxiv-cs-ro

arXiv:2605.28736v1 Announce Type: new Abstract: This study presents the first evaluation of general-purpose imitation learning for surgeon-robot collaborative assistance in open surgery, targeting sut

Implicit Regularization in Perturbed Deep Matrix Factorization: Spectral Conditions and Stability

arXiv:2605.28613v1 Announce Type: cross Abstract: This paper studies the stability of low-rank implicit regularization in perturbed deep matrix factorization, where the target matrix is corrupted by a

researcharxiv-cs-lg

Important context for latest OpenAI announcement. Especially (5:30): 'The model spit out a long transcript of an answer. Then a team of expe…

Important context for latest OpenAI announcement. Especially (5:30): 'The model spit out a long transcript of an answer. Then a team of expert mathematicians poured over this [transcript] and identifi

safetygary-marcus--x

Improving Evaluation of Recombination-based Cartesian Genetic Programming

arXiv:2605.28353v1 Announce Type: cross Abstract: Cartesian Genetic Programming has traditionally been using mutation as its main and often sole genetic operator to drive evolutionary search. Despite

Improving Requirements Classification with SMOTE-Tomek Preprocessing

arXiv:2501.06491v3 Announce Type: replace-cross Abstract: This study emphasizes the domain of requirements engineering by applying the SMOTE-Tomek preprocessing technique, combined with stratified K-f