AI News Today

arXiv:2603.19423v2 Announce Type: replace-cross Abstract: Large language model (LLM) agents increasingly rely on external tools (file operations, API calls, database transactions) to autonomously complete complex multi-step tasks. Practitioners deploy defense-trained models to protect against prompt injection attacks that manipulate agent behavior through malicious observations or retrieved content. We reveal a fundamental \textbf{capability-alignment paradox}: defense training designed to impro

arXiv:2604.08552v2 Announce Type: replace-cross Abstract: Scientific metadata are often incomplete and noncompliant with community standards, limiting dataset findability, interoperability, and reuse. Even when standard metadata reporting guidelines exist, they typically lack machine-actionable representations. Producing FAIR datasets requires encoding metadata standards as machine-actionable templates with rich field specifications and precise value constraints. Recent work has shown that LLMs

arXiv:2604.11556v2 Announce Type: replace-cross Abstract: LLM-assisted software development has become increasingly prevalent, and can generate large-scale systems, such as compilers. It becomes crucial to strengthen the correctness of the generated code. However, automated reasoning for large-scale systems remains challenging due to code complexity. Hoare logic offers an approach to decomposing a large system into smaller components and reasoning about them separately (i.e., compositional reaso

arXiv:2605.22748v2 Announce Type: replace-cross Abstract: Autonomous systems have achieved superhuman performance in isolation or simulation, yet they remain brittle in shared, dynamic real-world spaces. This failure stems from the dominant single-agent paradigm for physical applications, where other actors are ignored or treated as environmental noise, preventing effective coordination. Here we show that multi-agent reinforcement learning provides the essential safety scaffolding required for r

arXiv:2605.23733v3 Announce Type: replace-cross Abstract: Whole-body tracking (WBT) models have become a key foundation for humanoid robots, enabling them to imitate diverse motions with high fidelity. Training such models from scratch requires large-scale data and computation, making rapid deployment on new humanoid platforms costly. This raises a natural question: Can pretrained WBT models transfer across embodiments with minimal adaptation? To answer this question, we propose Any2Any, a parad

arXiv:2606.03090v2 Announce Type: replace-cross Abstract: The emergence of large language models (LLMs) has significantly accelerated recent research on LLM-based automatic grading (AG) systems. Benefiting from the strong instruction-following capabilities and broad prior knowledge of LLMs, educators can deploy AG systems across diverse tasks using only natural language rubrics while achieving satisfactory grading performance. Despite these advantages, new security concerns may also arise. In pa

arXiv:2606.05833v2 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) excel at 2D semantic understanding but lack intrinsic 3D awareness, resulting in representations that fail to maintain geometric and spatial consistency across video frames. Given the scarcity of large-scale 3D data, we present GeoVR, a novel framework that learns geometric representations using purely 2D video sequences. This approach effectively restructures the semantic latent space within MLLMs

arXiv:2606.12500v2 Announce Type: replace-cross Abstract: Traffic microsimulation combined with surrogate safety measures has increasingly been used as a proactive alternative to historical crash data for predicting crash frequency for current or planned road infrastructure designs. However, existing microsimulation-based safety studies have adopted simplified rule-based behaviour models, which reproduce traffic flow reasonably well but often fail to generate realistic conflict dynamics, limitin

arXiv:2606.16326v2 Announce Type: replace-cross Abstract: Paper A defines a time-consistent actuarial runtime that prices each side-effect-bearing action against a contractually fixed safe default and gates execution against a reserve budget. It treats the operator as passive. This paper makes the operator strategic. We characterise a five-attack space for autonomous AI-agent insurance contracts and prove when the actuarial runtime is gaming-resistant. Two attack surfaces -- post-toll safe-defau

arXiv:2606.18265v2 Announce Type: replace-cross Abstract: As human relationships with artificial intelligence systems become increasingly frequent and sustained, existing language and theory fail to accurately capture the nature of these affiliations. Common descriptors such as mutual understanding, connection, or friendship risk anthropomorphizing systems that lack subjective experience, while dominant frameworks tend to reduce AI to either a tool or a threat. In this paper, I introduce the con

arXiv:2606.18272v2 Announce Type: replace-cross Abstract: This paper presents an autonomous agentic resource negotiation framework designed to enable zero-touch network slicing in 6G architectures using Large Language Model (LLM) agents. While LLMs offer powerful reasoning capabilities, we demonstrate that such agents inherently suffer from anchoring bias, rigidly adhering to initial heuristic proposals and causing severe network over-provisioning. To systematically mitigate this cognitive bias,

arXiv:2606.18325v2 Announce Type: replace-cross Abstract: Enterprise intrusion response still depends on static playbooks and analyst-driven triage, creating delay between alert generation and containment. We present Agentra, a supervisable multi-agent Intrusion Response System (IRS) framework that converts alerts from IDS, EDR, and XDR platforms into structured incident response plans grounded in MITRE ATT&CK, MITRE D3FEND, and NIST CSF 2.0. Agentra decomposes response reasoning across role-sco

arXiv:2606.18611v2 Announce Type: replace-cross Abstract: We propose a parameter-efficient speech enhancement framework, Quaternion Conformer GAN (QC-GAN), which combines a Quaternion Conformer generator with MetricGAN-based training. The Hamilton product encodes the magnitude and phase via structured weight sharing, reducing the number of layer parameters while preserving their interdependencies. A metric-learning discriminator was employed to maximize perceptual quality by optimizing the appro

arXiv:2606.18996v2 Announce Type: replace-cross Abstract: Agents are increasingly deployed in document-intensive workflows where sensitive private information is not an edge case but a routine input, e.g., an agent booking a flight needs passport numbers. In such settings, the agent must use private information to complete tasks accurately while never exposing it in its responses, because it cannot verify who is actually at the keyboard. These two obligations are in fundamental tension. A model

Article URL: https://blog.curiousquail.com/spirit-crossings-ai-problem/ Comments URL: https://news.ycombinator.com/item?id=48606238 Points: 1 # Comments: 0

Article URL: https://github.com/agentark-ai/AgentArk Comments URL: https://news.ycombinator.com/item?id=48606186 Points: 2 # Comments: 0

Article URL: https://medium.com/@xavierbisits/how-to-learn-iraqi-arabic-bd468eff29e5 Comments URL: https://news.ycombinator.com/item?id=48606040 Points: 1 # Comments: 0

Robotic systems lack physics-aware training data for precise industrial assembly tasks involving force, torque, and tight-tolerance interactions, limiting real-world automation performance. Juakali is a physics-based data pipeline that makes robot datasets with force feedback accessible to automate assembly planning. It simulates fittings and fastening of mechanical parts from unstructured Computer-Aided Design (CAD) files while generating high quality structured datasets used to train Vision-La

CozoDB is an embedded relational-graph-vector database created by Ziyang Hu and the Cozo Project authors. It does Datalog, vector search (HNSW), and full-text search in a single embedded engine and was created with the vision to be the “Hippocampus for AI”. Unfortunately, the project went quiet after December 2024. KuzuDB, another embedded graph database with vector and full-text indexes similar to Cozo, was archived in October 2025. I decided to continue the vision by using Cozo to build a memo

Article URL: https://www.wsj.com/tech/ai/jane-street-ai-wall-street-bdfcc81a Comments URL: https://news.ycombinator.com/item?id=48605753 Points: 1 # Comments: 0

Could the Universe hold minds built from radically different stuff? ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://mojimoshi.com/ Comments URL: https://news.ycombinator.com/item?id=48605570 Points: 1 # Comments: 0

If LLM's are a file full of weights, does that mean somewhere out there lies a tiny SD card of an entire frontier model? What would you do if you found mythos laying on the ground? Obviously that's only half the puzzle, but how much would this SD card be worth? Could you simply copy paste it to 100's of other microSD cards and each one still be worth the same? Would it be considered a nuke in the wrong hands? Comments URL: https://news.ycombinator.com/item?id=48605536 Points: 2 # Comments: 0

Article URL: https://www.businessinsider.com/alphafold-john-jumper-leaves-google-deepmind-anthropic-demis-hassabis-nobel-2026-6 Comments URL: https://news.ycombinator.com/item?id=48605484 Points: 2 # Comments: 0

Article URL: https://entertainmentindustry.ai Comments URL: https://news.ycombinator.com/item?id=48605422 Points: 1 # Comments: 0

<p>A pragmatic backup and restore workflow for moving Codex Desktop local state, threads, plugins, and skills between Macs.</p> <p>Am an Open AI Partner and I'm in the middle of switching Macs from M1 Air to M4 Air. What. a. pain. Especially in the enterprise world where you cannot just simply migrate everything from one Mac to another. You need to set it up as new. So here we go...</p> <p>I didn't want too much ceremony moving my, err, precious, Codex Desktop setup from one Mac to another and w

<pre style='white-space:pre-wrap;width:81ex'>fix(test): guard claude usage debug args</pre>

<p>A while back I wrote a post called <em>"<a href="https://dev.to/_vjk/i-made-claude-code-think-before-it-codes-heres-the-prompt-bf">I Made Claude Code Think Before It Codes. Here's the Prompt.</a>"</em> The premise struck a nerve: the problem with AI coding assistants was never <em>intelligence</em>. These models can write a working function faster than you can describe it. The problem was <strong>process</strong>. Left to its own devices, Claude Code behaves like a brilliant junior developer

<p>After Day 1 wrapped, I was waiting for Day 2 — Voice Agent day.</p> <p>Honestly, Voice Agent was the piece I was most nervous about in this whole system.</p> <p>Because the other agents — write emails, look up data, run routing — at worst they write something mediocre, get blocked by Tone Coach, and rewrite once. But Voice Agent actually picks up the phone and calls a real person. The other side picks up, hears Sarah's voice, hears the script I wrote, gets the routing I designed.</p> <p>Any m

Article URL: https://surfingcomplexity.blog/2026/06/19/i-am-dreading-our-llm-written-incident-report-future/ Comments URL: https://news.ycombinator.com/item?id=48605136 Points: 3 # Comments: 0

Article URL: https://metalearn.substack.com/p/i-solved-my-mystery-fatigue-with-ai Comments URL: https://news.ycombinator.com/item?id=48605117 Points: 21 # Comments: 5

French serial entrepreneur and open-source legend Jean-Baptiste Kempf has been building Kyber, an infrastructure layer to control remote devices in real time.

<p>A year ago I was the guy rolling his eyes at the AI demos.</p> <p>I tried to use it back then to score sales leads, and I couldn’t get it to reliably do something as basic as flag every company with more than 25 employees. A plain “x > 25” was beyond it.</p> <p>That wasn’t the 10x revolution everyone kept promising. Not yet.</p> <p>Things have changed fast.</p> <p>A year later, I just spent two weeks and about $4,000 in AI tokens rebuilding our entire website. I did it with what I’ve been cal

Article URL: https://www.agentcard.sh Comments URL: https://news.ycombinator.com/item?id=48605055 Points: 1 # Comments: 0

<p>AI coding agents are getting good enough that they no longer feel like autocomplete.</p> <p>Tools like Claude Code, Codex, OpenCode, Cursor, Cline, and other agentic coding systems can read files, modify code, run commands, call tools, inspect errors, and continue working across multiple turns.</p> <p>That is useful. It is also increasingly opaque.</p> <p>When you ask an agent to fix a bug, you usually see the final answer and maybe a few tool calls in the terminal. But you often do not know:

<p>The story everyone is reading wrong is that <strong>AI's real bottleneck is electricity, not chips</strong>. On <strong>June 18, 2026</strong>, the US <strong>Federal Energy Regulatory Commission (FERC)</strong> ordered six major grid operators to give data centers a fast lane onto the grid, per <a href="https://techcrunch.com/2026/06/18/ai-data-centers-just-got-a-government-mandated-fast-lane-to-the-grid/" rel="noopener noreferrer">TechCrunch</a>. What it did not do is add a single megawatt

<h1> How Machine Learning Detects Fraud: A Practical Breakdown </h1> <p>Machine learning detects fraud by learning the patterns of past fraudulent<br> transactions and flagging new transactions that match those patterns —<br> combining models trained on known fraud cases with anomaly-detection methods<br> that catch fraud patterns no one has seen before. Most production fraud<br> systems use both approaches together, not one or the other.</p> <p>Here's how that actually works, and what makes fra

<h1> How AI Will Shape the Technology Industry in 2027 </h1> <p>We're roughly 6 months out from 2027, and the signals are already converging: AI is not coming — it has arrived, and the next wave will be fundamentally different from everything that came before it. For developers and tech professionals, 2027 isn't a distant horizon. It's the next major inflection point to prepare for now.</p> <p>Here's what the research, analysts, and industry leaders are saying about what's ahead.</p> <h2> From G

<p>AI isn't coming for your job.</p> <p>It already showed up, merged its first PR, and doesn't need a code review.</p> <p>The question developers keep dancing around — but rarely say out loud — is this:</p> <blockquote> <p><strong>If GitHub Copilot, Cursor, and Claude can do what a junior dev does in a fraction of the time, what happens to junior devs?</strong></p> </blockquote> <p>And more uncomfortably: what happens to <em>mid-level</em> devs in three years?</p> <h2> The Uncomfortable Data Poi