AI News Today

arXiv:2607.19830v1 Announce Type: new Abstract: Hypergraph-based RAG systems surpass traditional graph-based approaches by organizing complex n-ary atomic facts among entities, rather than relying solely on binary relationships. Despite the advancements in multimodal large language models (MLLMs) with enhanced visual capabilities, current hypergraph-based RAG frameworks predominantly restrict knowledge retrieval and reconstruction to a unimodal, text-centric paradigm. This limitation prevents th

arXiv:2607.19834v1 Announce Type: new Abstract: With the wide application of large language models (LLMs) in real-world scenarios, the value implication of their outputs is crucial. However, existing evaluation benchmarks suffer from insufficient coverage of value dilemmas in daily scenarios involving multiple value conflicts and simplistic evaluation formalisms that fail to assess LLMs' value alignment. To address these issues, we propose D2VBench, a value alignment benchmark comprising 10,000

arXiv:2607.19848v1 Announce Type: new Abstract: There is growing evidence that data diversity is crucial for developing fair and robust NLP models. However, current approaches to measure diversity remain inconsistent and fragmented: While there exist a number of tools for measuring the lexical diversity of texts, researchers lack standardized tools for quantifying diversity based on embeddings. Embedding-based diversity measures are highly flexible: They work with any embedding model and any dat

arXiv:2607.19932v1 Announce Type: new Abstract: Spoken language models (SLMs) enable natural human-computer interaction, but their reasoning ability still lags behind that of text-based large language models, especially on spoken mathematical question answering tasks. One important reason is that SLMs reason over purely verbalized mathematical expressions, which are harder to interpret than symbolic text. However, directly transferring text-based reasoning to SLMs is nontrivial due to architectu

arXiv:2607.19956v1 Announce Type: new Abstract: Knowledge distillation (KD) is a standard approach for compressing sequence-to-sequence models, but its per-sample effects are rarely examined. On the BanSum Bangla summarization benchmark, we find that standard KD improves ROUGE-L by only +0.0003 over a cross-entropy baseline, and that approximately 51.3% of training samples are estimated to actively harm student validation loss under standard KD. We propose two complementary reliability-aware dis

arXiv:2607.19992v1 Announce Type: new Abstract: tiny_schiller closes the small-language-model prototyping, fine-tuning, education, and research gap for German literary text, providing a single-file, drop-in counterpart to Karpathy's tiny_shakespeare. The available German literary corpora are larger and richer, but require parser engineering before a single line of training or fine-tuning code can run. tiny_schiller is a 2.07-megabyte single file of eleven public-domain Schiller dramas, sourced f

arXiv:2607.20009v1 Announce Type: new Abstract: This paper presents the second edition of the TalentCLEF Challenge, which will run as an evaluation lab as part of CLEF 2026. The aim of TalentCLEF is to promote the development of systems and methods that use Natural Language Processing (NLP) in the field of Human Capital Management (HCM), fostering approaches that ensure fairness in results, operate across multiple languages, and adapt to diverse industries. To this end, TalentCLEF establishes pu

arXiv:2607.20062v1 Announce Type: new Abstract: We present Solar Open 2, a 250B-A15B Mixture-of-Experts language model built for long-horizon agentic tasks, scaled up from Solar Open 1 (Solar Open 100B). To hold entire agent trajectories in a single context, Solar Open 2 reaches a 1M-token window through a hybrid attention stack that interleaves one softmax layer among every three linear-attention layers, using no positional encoding and a gated delta rule extended to negative eigenvalues. To tr

arXiv:2607.20082v1 Announce Type: new Abstract: Language models are increasingly asked to self-report, informing safety evaluations, public understanding, and model-welfare debates. Yet their reports are elicited with human questionnaires never validated for models or ad hoc prompts of unknown reliability. We propose the first language-model-specific psychometric theory: a two-process theory of machine self-report. Self-description jointly reflects persona installation, through which post-traini

arXiv:2607.20090v1 Announce Type: new Abstract: Retrieval-augmented large language models frequently face contexts that interleave useful evidence with misleading statements or instruction-like content. Blanket refusal discards valid evidence, whereas uncritical adoption yields incorrect or unsafe answers. The ability to selectively adopt relevant information while rejecting deceptive or harmful content is therefore critical for reliable deployment in real-world retrieval settings. We introduce

arXiv:2607.20115v1 Announce Type: new Abstract: Large language models (LLMs) are known to be sensitive to prompt and input formulations. However, existing studies have focused on lexical realization and largely ignored constructional choice. This paper studies whether linguistic construction can systematically shift LLM decisions and where these shifts can be causally localized inside the model. We use political stance judgment as a meaning-sensitive case study and extend an English political st

<p> The 1st Firewall for AI Agents </p> <p> <a href="https://www.producthunt.com/products/hol-guard?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/1204113?app_id=339">Link</a> </p>

NVIDIA founder and CEO Jensen Huang today visited the Naval Postgraduate School in Monterey, California, to commission an NVIDIA DGX GB300 system — bringing one of the world’s most powerful AI platforms fully online for the students, researchers and faculty at the U.S. military’s flagship graduate university. “Our nation depends on our men and women […]

<p> Run GPT-5.6 on OpenClaw & Hermes safely on your Mac </p> <p> <a href="https://www.producthunt.com/products/plow-mac-app-safely-run-gpt-5-6-agents?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/1204092?app_id=339">Link</a> </p>

Article URL: https://github.com/PxyUp/fitter Comments URL: https://news.ycombinator.com/item?id=49015561 Points: 1 # Comments: 0

Article URL: https://www.localmaxxing.com/en Comments URL: https://news.ycombinator.com/item?id=49015424 Points: 1 # Comments: 2

You can now add GitHub tools to your eve agent as an extension . Add the package, drop one file in agent/extensions/ , and your agent gets every tool with Vercel Connect auth, presets, and approval rules built in. Install @github-tools/eve-extension : Then register it from a file in agent/extensions/ : Connector-backed auth: Pass a Vercel Connect connector and the extension mints short-lived, scoped GitHub tokens at runtime. Presets scope the toolset: code-review , issue-triage , repo-explorer ,

Ling 3.0 Flash from Ant Group is now available on AI Gateway. The model is free to use for the next three weeks, through August 3rd. Ling 3.0 Flash is a Mixture-of-Experts model with 124B total parameters and about 5.1B active per token. It has a 256K token context window and runs in thinking and non-thinking modes. Ling 3.0 Flash is built for token-efficient agentic inference at production scale, doing more work within tighter token, latency, and cost budgets across multi-step agent runs. The m

The Vercel MCP server can now deploy code directly to a new or existing project. When your AI assistant finishes building something, it can ship it to Vercel and hand back a shareable URL without leaving the chat. Point the deploy_to_vercel tool at your files and Vercel creates the project, detects the framework, installs dependencies, and builds. You get a URL you can open and share while the build finishes in the background. To get started, connect the Vercel MCP server to Claude, Cursor, or a

Health in ChatGPT now lets eligible U.S. users securely connect medical records and Apple Health to get more personalized insights and better understand their health.

<blockquote cite="https://twitter.com/tqbf/status/2080045032162173329"><p>I genuinely believe that if you took an open weights model from 2025 and built a pentest harness for it, it could do this kind of sandbox escape and scan/hack in most networks. This is only surprising because you assume OpenAI has sounder sandboxes.</p></blockquote> <p class="cite">— <a href="https://twitter.com/tqbf/status/2080045032162173329">Thomas Ptacek</a>, doesn't think <a href="https://simonwillison.net/2026/Jul/22

<p>This story is wild. The short version: OpenAI were running a cybersecurity test against an unreleased model, with the model's guardrail features turned off. Rather than solve the test, the model broke its way out of OpenAI's sandbox, then found exploits to break <em>in</em> to Hugging Face, all so it could cheat on the test by stealing the answers.</p> <p>Along the way it helped make the strongest case yet for how the imbalance of model availability is hurting our ability to secure our softwa

<p> Full-Stack Platform for Training Small Language Models </p> <p> <a href="https://www.producthunt.com/products/freesolo-flash?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/1204056?app_id=339">Link</a> </p>

After IBM's stock crashed last week on warnings of poor mainframe sales, the CEO explained that AI wrecked corporate hardware budget, temporarily.

Closed models with guardrails can still cause harm, but may also not be able to fix problems they caused

<p><strong><a href="https://dylancastillo.co/posts/pelicanmaxxing.html">Are AI labs pelicanmaxxing?</a></strong></p> Excellent piece of work by Dylan Castillo, who took a deep-dive into the frequently pondered question of whether the AI labs have been deliberately training models to draw pelicans riding bicycles in response to my <a href="https://simonwillison.net/tags/pelican-riding-a-bicycle/">deeply unscientific benchmark</a>.</p> <p>I've been randomly spot-checking this in the past by testin

Article URL: https://arxiv.org/abs/2607.18250 Comments URL: https://news.ycombinator.com/item?id=49014537 Points: 4 # Comments: 0

<!-- SC_OFF --><div class="md"><p>We spent the last months consolidating seven separate sequence classifiers into one multi-head model, our apex model, so to speak, and since the weights are now public, I wanted to share what worked and what surprised us.</p> <p><strong>Setup:</strong> a shared mmBERT-small encoder with seven task heads, binary injection (BCE), document class (7-way), tool type (14-way), tool operation (6-way), tool data-flow tags (3× BCE, multi-label), intent routing (5-way), a

Here are the stats. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

When Hugging Face got hit last week, co-founder Clement Delangue suspected a frontier lab, given the agent's sophistication. He was right. Delangue said on X that after a day working with OpenAI he strongly believed there was no malicious intent and that it was mind-blowing the whole thing had happened autonomously. The two OpenAI models that broke into Hugging Face last week did not breach it through malice or superintelligence. They breached it through credentials and permissions they should n

<p> A research engine for your agent </p> <p> <a href="https://www.producthunt.com/products/webhound?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/1204010?app_id=339">Link</a> </p>

<p> Build a website from Google Business, Instagram, or Facebook </p> <p> <a href="https://www.producthunt.com/products/repaint-ai-website-builder?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/1204009?app_id=339">Link</a> </p>

Google's cloud business is thriving, as companies adopting its AI and AI infrastructure services help the tech giant to report record profits.

Neill Blomkamp's AI-generated short film, Nightborne, is facing backlash for being derivative and boring.

Article URL: https://www.roleframe.ai/blog/why-ai-resume-sounds-generic Comments URL: https://news.ycombinator.com/item?id=49013985 Points: 2 # Comments: 0

<p> Undetectable AI copilot for live meetings </p> <p> <a href="https://www.producthunt.com/products/mufal?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/1204001?app_id=339">Link</a> </p>

Article URL: https://twitter.com/joefioti/status/2077430481809989662 Comments URL: https://news.ycombinator.com/item?id=49013890 Points: 1 # Comments: 0

Article URL: https://github.com/ykushch/agsig Comments URL: https://news.ycombinator.com/item?id=49013862 Points: 1 # Comments: 0