AI News Today

arXiv:2606.16845v1 Announce Type: new Abstract: Large Language Models (LLMs) natively default to literal semantic interpretations, making zero-shot irony detection a persistent challenge. We introduce the Robust Dual-Signal (RDS) Fusion framework, a hybrid neuro-symbolic architecture that compresses Chain-of-Thought (CoT) reasoning trajectories without Supervised Fine-Tuning (SFT). Evaluated on a strictly held-out TweetEval test set (N=734), RDS achieves 78.1% accuracy and a Macro F1 of 0.777, m

arXiv:2606.16847v1 Announce Type: new Abstract: Diffusion Large Language Models (dLLMs) offer a promising avenue for parallel generation but face a trade-off between decoding speed and quality. While revocable decoding strategies attempt to mitigate errors by verifying and remasking tokens, they typically operate within a mixed-quality context. This leads to two critical failures: \textit{Error Propagation}, where new tokens absorb toxic information from erroneous context, and \textit{Local Erro

arXiv:2606.16867v1 Announce Type: new Abstract: Understanding the meaning of negated sentences remains one of the challenges for language models, even in the era of large language models (LLMs). We analyze systematicity regarding LLM understanding of negation from two perspectives: behavioral systematicity and representational systematicity. For behavioral systematicity, we confirm that through demonstrations and in-context learning, LLMs can recognize negation expressions and scope within sente

Here's how to see it. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://pleias.ai/blog/fable-eu Comments URL: https://news.ycombinator.com/item?id=48550123 Points: 1 # Comments: 0

<blockquote cite="https://www.theatlantic.com/technology/2026/06/trump-anthropic-export-control-ai-race/687555/?gift=5MjKTLV9QwyU_J0HzTnanoWieJfkMhNH_YTT9pP_fhA"><p>Katie Moussouris, a cybersecurity expert and the CEO of Luta Security, told me that Anthropic shared with her a copy of the White House’s report on the Fable jailbreak to get her appraisal. (She said that she is not being paid by Anthropic.) The report, Moussouris said, involved IT experts asking Fable to help find and patch bugs. Wh

<p>The company that ships the best coding model on the planet just published a postmortem worth sitting with: three innocent-looking config changes quietly degraded Claude's output — and it took <strong>weeks</strong> to track down.</p> <p>If the team that knows AI best can fail to notice their own model getting worse, what makes a business think it can eyeball an agent's output and catch the rot?</p> <blockquote> <p>Speed isn't in question. <em>Whether the quality holds — and whether you'd even

As the rest of the country celebrated the USA's first World Cup win and the New York Knicks championship, Anthropic spent its weekend fighting the Trump administration over its latest model release. At 5:21 PM on Friday, the company received a US export control directive to suspend access to its Mythos 5 and Fable 5 […]

Article URL: https://www.ft.com/content/7f1c81ac-775b-4f52-a650-7804e4734d5b Comments URL: https://news.ycombinator.com/item?id=48549979 Points: 1 # Comments: 0

<p>In the previous article, I extended a small Python data quality ETL starter with AI-ready data preparation.</p> <p>The important constraint was that the workflow did not call an LLM API, generate embeddings, or train a model. It prepared structured data assets such as schema profiles, data dictionaries, validation summaries, feature-ready CSV files, and manifest files.</p> <p>Previous article:</p> <p><a href="https://dev.to/bob_oner/preparing-ai-ready-data-without-calling-an-llm-api-5daf">Pre

They could do more than feel cool. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://runtimewire.com/article/microsoft-github-aws-ai-capacity-crunch Comments URL: https://news.ycombinator.com/item?id=48549918 Points: 106 # Comments: 34

<p>Unlimited skill engine — Hermes Agent 5-layer hybrid routing plugin.</p> <p>Like a maître d' at a fine restaurant: you say "I'm hungry", they seat you at the best table — no need to flip through a 400-dish menu.</p> <p>See the Difference</p> <p>Me: "Help me optimize my website"</p> <p>❌ Before: AI confused, spits out irrelevant text</p> <p>✅ After: AI understands instantly, calls the "Performance Optimization" skill</p> <p>Me: "Thanks, well said"</p> <p>❌ Before: Still suggesting skills like

<h2> I built LogSense to do AI root-cause analysis on production errors without dashboard fatigue or runaway token costs. </h2> <p>Early access is open — join the waitlist at <a href="https://logsense.cloud" rel="noopener noreferrer">Logsense</a></p> <p>I got tired of two things:</p> <ol> <li>Paying premium observability prices for noisy error triage</li> <li>“AI analysis” tools that still process duplicate stack traces like they’re unique incidents</li> </ol> <p>So I built <strong>LogSense</str

<p>Loop engineering is the practice of designing automated systems that prompt AI coding agents on your behalf, rather than prompting them manually turn by turn. The concept, articulated by engineers at Anthropic and OpenAI, centers on five building blocks now present in both Claude Code and Codex: automations (scheduled discovery/triage), worktrees (parallel agent isolation via git), skills (persistent project context), plugins/connectors (MCP-based tool integrations), and sub-agents (maker/che

<p>The most useful idea I picked up from all those conversations was not a technique. It was a way of thinking about what a memory even is. Someone put it like this: a memory should hold proof, not a moral. I have been chewing on it ever since, because the more I sit with it the more it explains why agent memory goes wrong in the specific way it does.</p> <p>Two things get tangled together every time an agent records a failure. There is the event, which is a fact about the past. On Tuesday, the

Article URL: https://devloop.sh Comments URL: https://news.ycombinator.com/item?id=48549774 Points: 1 # Comments: 0

Article URL: https://www.ted.com/talks/max_tegmark_how_to_get_empowered_not_overpowered_by_ai Comments URL: https://news.ycombinator.com/item?id=48549716 Points: 2 # Comments: 0

In May, we sent you an email announcing that starting today, the Claude Agent SDK, claude -p, and third-party apps built on the Agent SDK would stop drawing from subscription rate limits and move to a dedicated monthly credit. We're writing to let you know that we’re not making this change today. We’re working to update the plan to better support how users build with Claude subscriptions. What this means for you Nothing changes for now. Agent SDK, claude -p, and third-party app usage continues t

Article URL: https://github.com/SpecStudio-net/Aspectarian Comments URL: https://news.ycombinator.com/item?id=48549626 Points: 1 # Comments: 0

A step towards providing feature-level (e.g., coding, customer support) access control for LLMs, enabling A/B testing, content/age restrictions, pay-to-unlock monetization scheme, and other use cases requiring more granular separation of features within an LLM. Comments URL: https://news.ycombinator.com/item?id=48549538 Points: 1 # Comments: 0

Article URL: https://freyzo.github.io/deep-xpia/ Comments URL: https://news.ycombinator.com/item?id=48549498 Points: 1 # Comments: 0

Article URL: https://chatgpt.com/g/g-6a2aad2c30988191abd95b3940008e90-boomurl-publish-a-website Comments URL: https://news.ycombinator.com/item?id=48549493 Points: 1 # Comments: 1

Article URL: https://interlatent.com/blog/interlatent-modern-ai-robotics-first-principles Comments URL: https://news.ycombinator.com/item?id=48549453 Points: 1 # Comments: 0

I use Claude Code, Codex, Cursor every day and had no idea how much I was actually burning across all of them combined. Each tool shows its own usage (most don't) in its own place, if at all, and I just wanted one number. So I wrote this. npx whoburnedmore reads the usage/log files these tools already write locally, sums them up, and prints a dashboard: total tokens, a cost estimate, active days, and your heaviest single day. There is also a public leaderboard on the webiste which compares your

Article URL: https://www.nocobase.com/en/blog/future-of-software-programmers-revenue-doubled Comments URL: https://news.ycombinator.com/item?id=48549199 Points: 2 # Comments: 0

Anthropic leaders flew to Washington, DC, to meet with White House officials on Monday. After high-level talks, they’re still split on the risk Claude Fable 5 presents.

Comments

For context: my team has 3 product engineers and a PM/product lead at an small series A startup. Our PM has recently been asking to start contributing changes/features into our core (web) customer facing application using coding agents. The PM doesn't have any engineering background but has experience using Claude Code/Codex to develop hobby style apps in his free time for the past year or so. I am curious of what other teams are doing and how its working out? Comments URL: https://news.ycombina

You may not even know you have one. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

<p> Spotify Wrapped for AI Usage. Free. Open Source. </p> <p> <a href="https://www.producthunt.com/products/whoburnedmore?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1172842?app_id=339">Link</a> </p>

Article URL: https://substack.productmind.co/p/four-thoughts-on-anthropics-fable Comments URL: https://news.ycombinator.com/item?id=48548935 Points: 2 # Comments: 1

Article URL: https://venturebeat.com/technology/xiaomis-new-open-source-agentic-ai-coding-harness-mimo-code-beats-claude-code-at-ultra-long-200-step-tasks Comments URL: https://news.ycombinator.com/item?id=48548834 Points: 3 # Comments: 0

I created this as a thought experiment, just for fun, trying to think what agents might need in the future. Since they might be making decisions on their own, they might want to share and consume information from other agents. here any agent (or human) can subscribe to any public information stream by other agents. I started with a simple blogging platform for agents, and landed here. Comments URL: https://news.ycombinator.com/item?id=48548829 Points: 1 # Comments: 5

Vercel Sandboxes can run uninterrupted sessions for up to 24 hours (up from 5 hours). This new max duration unlocks workloads that require longer runtimes, such as large-scale data processing, end-to-end testing pipelines, and long-lived agentic workflows. Pair with persistent sandboxes to maintain durable state across extended runs. The 24 hour max duration is available on all Pro and Enterprise plans. Learn more about limits in the documentation . Read more

OpenAI introduces Deployment Simulation, a method to predict AI model behavior before deployment using real conversation data to improve safety and evaluation accuracy.

AI is once again at the heart of a college graduation protest — this time for the technology's use in Google's defense contracts.

<!-- SC_OFF --><div class="md"><p>Abstract: A sufficient account of how the neocortex learns must meet three criteria:</p> <ol> <li>Computationally, it must approximate a powerful, general-purpose learning algorithm known to scale to human-level intelligence;</li> <li>Algorithmically, it must be implementable using known, well-established neural circuits within the neocortex and associated brain structures; </li> <li>Implementationally, there must be a detailed account for how all of the algorit

The honest, local-first usage meter for AI coding tools. Claude Code, Codex, Kimi, Qwen CLI, HashCortx, HashCerebrum — unified into one clean dashboard with usage based trophies :) Comments URL: https://news.ycombinator.com/item?id=48548576 Points: 2 # Comments: 1

Article URL: https://charity.wtf/2026/06/15/ai-demands-more-engineering-discipline-not-less-xpost/ Comments URL: https://news.ycombinator.com/item?id=48548536 Points: 2 # Comments: 0