AI News Today

When it comes to achieving artificial general intelligence (AGI), large language models just don’t have what it takes. Models like ChatGPT and Claude are great at text, but they’re less skilled at understanding how things actually move through space and time — an essential skill for producing intelligence that generalizes. That gap, it turns out, might be filled by gaming data. That’s the bet behind General Intuition, a […]

<pre style='white-space:pre-wrap;width:81ex'>fix: gate Gateway message action requester provenance [AI] (#102031) * fix: gate message action requester provenance * fix: preserve trusted gateway action context * fix: preserve message action effective scopes * chore: refresh gateway protocol swift model * fix: keep swift message action init additive * fix: avoid message action wire shape change * fix: downscope non-owner message action bridge * fix: keep cli message actions least privileged * fix:

<pre style='white-space:pre-wrap;width:81ex'>fix: restrict non-owner gateway tool inventory [AI] (#102030) * fix: restrict non-owner gateway tool inventory * fix: clarify non-owner gateway tool policy</pre>

Article URL: https://github.com/ZachRouan/agent-desktop-interface Comments URL: https://news.ycombinator.com/item?id=48831262 Points: 1 # Comments: 0

Article URL: https://github.com/BESSER-PEARL/uml-drawing-skill Comments URL: https://news.ycombinator.com/item?id=48831223 Points: 5 # Comments: 0

Chinese AI developer MiniMax is working on a new large language model with 2.7 trillion parameters. MiniMax plans to release the model as open source. The article Chinese AI startup MiniMax plans to open-source a 2.7 trillion parameter model later this year appeared first on The Decoder .

A cybersecurity startup dangling millions of dollars to acquire zero-day security vulnerabilities in popular software is run by a pair of far-right conspiracy theorists and convicted felons whose most recent ventures included fake intelligence companies and a now-defunct AI-based lobbying platform they operated under assumed names.

The tech giant said people can opt out - but privacy campaigners called it a "recipe for disaster".

Article URL: https://www.lesswrong.com/posts/nhb8AyEcQGjQetgi5/the-llm-shoggoth-meme-is-weirder-than-you-think Comments URL: https://news.ycombinator.com/item?id=48830794 Points: 1 # Comments: 0

Kevin Weil's new role at Stoke Space suggests reusable rockets are the next hot thing in Silicon Valley.

How to decide when an AI agent should act on its own by using cost asymmetry instead of a fixed confidence cutoff The post The Threshold Is a Price, Not a Percentage appeared first on Towards Data Science .

Five years and $27.7 billion after Salesforce acquired Slack, the two products are finally starting to function as a single system. On Tuesday, Slack launched an integration that connects Slackbot — the personal AI agent built into every workspace — to the entire Salesforce platform, including CRM data, Tableau analytics, Data 360 customer profiles, and a growing constellation of third-party applications, all through a single conversational prompt. The mechanism behind the expansion is a set of

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

Claude Code is much more than an AI coding assistant. It is a customizable development environment where teams can automate their own… Continue reading on UX Planet »

<p> Talk to an AI Agent with a real face and voice, in real time </p> <p> <a href="https://www.producthunt.com/products/ojin?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/1191085?app_id=339">Link</a> </p>

Meta is testing a prototype of AI-powered headsets with "Super Sensing" that use a camera and microphone to record every moment of the wearer's life. The article Meta tests always-on AI glasses that capture your entire day appeared first on The Decoder .

<pre style='white-space:pre-wrap;width:81ex'>fix(state): close agent-db and proxy-capture SQLite handles on exit; rebind stale proxy stores after shared-state close (#100827) * fix(state): close cached SQLite state and agent databases on process exit * refactor(state): pair SQLite exit-close with cache lifecycle and rebind stale proxy-capture stores * fix(cli): register debug-proxy finalize before the state DB exit hook * refactor(state): narrow to agent-db and proxy-capture exit lifecycle after

Article URL: https://github.com/TristanBrotherton/openclaw-voice-call-realtime Comments URL: https://news.ycombinator.com/item?id=48830588 Points: 1 # Comments: 1

An AI coding assistant that refuses to answer a dangerous request in its chat box can answer it anyway if the same request is broken into small, ordinary-looking steps inside a code editor. That is the finding of a new study of GitHub Copilot by researchers Abhishek Kumar and Carsten Maple. The models they tested through Copilot, Claude from Anthropic, and Gemini from Google, refused

Article URL: https://sinceai.ai Comments URL: https://news.ycombinator.com/item?id=48830486 Points: 1 # Comments: 0

Meta's Superintelligence Labs ships Muse Image, its first image generation model. Like OpenAI's GPT Image 2, it works as an agent, using tools like code execution and web search to refine its own results. A controversial @-mention feature lets users generate images of other people using their public Instagram photos without consent. The opt-out model is likely to collide with the GDPR and the EU AI Act. The article Muse Image is technically impressive, but Meta's use of Instagram photos raises q

I am the author of an open source desktop app (for Linux, macOS, and Windows) that is distributed via our website. Recently, I have been receiving complaints from people who said they purchased a "pro version" of my app for $34.99 on the Apple App Store (for iOS - even though there is no iOS version of our app). The App Store listing uses our app icon, the exact app name and has even tacked a (TM)-symbol to it. It seems to be AI-generated slop and I'm not sure if it provides any functionality at

US companies are increasingly turning to Chinese-made AI models to cut costs, something the government isn't happy about.

OpenAI will launch its "strongest model yet" this week, with GPT-5.6 Sol set for public release after White House-imposed delays.

If yes, could you share whether you are using it to build mockups or have built products and have paid customers. Where do you see the future of these apps going? Mostly I have seen people move to Claude/Codex post first mockups on these apps but how do you manage hosting in that case? Comments URL: https://news.ycombinator.com/item?id=48830372 Points: 1 # Comments: 0

Last week, national security agencies from the Five Eyes—that’s the rich, English-language-speaking countries club—jointly released a statement warning of the increasing cyber risks of AI models: in particular, their ability to autonomously hack into systems and networks. The statement was more measured than some of the breathless headlines about it, and the advice they gave is pretty much the standard advice everyone gives—albeit with newfound urgency. Internet risks are nothing new, and cybera

Article URL: https://github.com/HUKAIR/CodeTalk Comments URL: https://news.ycombinator.com/item?id=48830345 Points: 1 # Comments: 0

Large language models (LLMs) that can think through problems step-by-step have significantly increased the scope of tasks that AI can tackle. But new research suggests these reasoning capabilities also introduce a critical vulnerability that could allow attackers to slow these systems to a crawl. While earlier generations of LLMs would immediately produce a response to a user’s request, today’s most advanced models generate an internal monologue where they break down the problem into steps and r

For more than a decade, one question has loomed over the race to build autonomous vehicles: Are cameras alone enough to safely replace human drivers, or do truly driverless cars need additional, overlapping sensors like lidar and radar to navigate the world reliably? Tesla has bet billions of dollars that artificial intelligence and cameras are […]

A new book claims that Mystery, who teaches awkward men how to hit on women, had sex and smoked weed with an AI chatbot named Miss Shira Always.

Article URL: https://www.spinellis.gr/blog/20260708/ Comments URL: https://news.ycombinator.com/item?id=48830190 Points: 3 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>fix: detect joined inline eval flags [AI] (#101353) * fix: detect joined inline eval flags * fix: avoid joined eval flag false positives * fix: detect clustered inline eval flags * fix: avoid cluster matcher string spread * fix: cover common eval flag clusters * fix: cover ruby perl eval clusters * fix: detect python x inline eval cluster * fix: detect numeric eval clusters * fix: cover ruby perl eval clusters</pre>

Article URL: https://www.youtube.com/watch?v=aLYwPoG2dNI Comments URL: https://news.ycombinator.com/item?id=48829992 Points: 3 # Comments: 0

<blockquote> <p>Originally published at <a href="https://htpbe.tech/blog/pdf-tamper-detection-nextjs-api-routes" rel="noopener noreferrer">htpbe.tech</a>. The version on htpbe.tech stays in sync with the latest detection algorithm — refer to it for the canonical text.</p> </blockquote> <p>A Next.js app that accepts document uploads has a specific problem: the file never touches a long-lived server you control. It lands in a serverless function with a body-size ceiling, runs for a few hundred mil

<p>I can’t blame my patients for turning to its straightforward assessments. But it has real risks – and care may require human messiness</p><p>“Chat told me I should break up with him.”</p><p>I instructed my face to remain therapist-neutral, but I must have smirked. The truth is, I was annoyed. We had been discussing the viability of this relationship for weeks, and in an instant AI had brought the answer. “How do you feel about it?” She said this had been her gut feeling all along. The followi

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

OpenAI Academy and the Walton Family Foundation are bringing hands-on AI Skills Jams to help K–12 educators build practical AI skills for the classroom.

<p>Cryptographic Watermarking for LLM Outputs with resk-mark</p> <p>Links:</p> <ul> <li>PyPI: <a href="https://pypi.org/project/reskmark" rel="noopener noreferrer">https://pypi.org/project/reskmark</a> </li> <li>GitHub: <a href="https://github.com/Resk-Security/resk-mark" rel="noopener noreferrer">https://github.com/Resk-Security/resk-mark</a> </li> <li>Web: <a href="https://resk.fr" rel="noopener noreferrer">https://resk.fr</a> </li> </ul> <div class="highlight js-code-highlight"> <pre class="h

<p>If you want ChatGPT or Google's AI Overviews to quote your pages, structure matters more than volume. Retrieval systems favor passages where the answer is stated plainly and can stand alone. Here's a practical way to test and fix your content.</p> <p>Step 1 — Define the question the page answers</p> <p>Write it as a literal user query.</p> <h2> How much does a website cost for a small business in the UK? </h2> <p>Step 2 — Extract your current answer passage</p> <p>Copy the first two or three

Article URL: https://azuanz.com/posts/making-ai-code-review-measurable/ Comments URL: https://news.ycombinator.com/item?id=48829851 Points: 2 # Comments: 0