AI News Today

<!-- SC_OFF --><div class="md"><p>I’m pretty annoyed. We’re a small AI startup building a real-time coding agent. Our p95 latency requirements are tight (and self imposed, but thats the product). We need sustained high-throughput inference with ~1-2k tokens/second. Been on the Cerebras waitlist for months trying to get API access. We’re not doing training so don’t need a warehouse of H100s. We need fast, high-throughput ASIC inference for a specific production workload. Cerebras’ just went publi

The rise of small language models The post How to Choose Between Small and Frontier Models appeared first on Towards Data Science .

The car-maker found AI quality checks failed to match the skill of veteran technicians.

I was tired of asking my claude code to reference my codex chats to get references to what decisions it made and why ; so I built Reference MCP It, whenever prompted establishes sessions to get direct access - been using it on my system for a bit and was super helpful so I made a repo :) Would love feedback! Comments URL: https://news.ycombinator.com/item?id=48718055 Points: 3 # Comments: 0

Article URL: https://aicompiler.dev Comments URL: https://news.ycombinator.com/item?id=48718031 Points: 2 # Comments: 0

Today’s encrypted data, such as credentials, may no longer remain confidential in the future because the public-key cryptography protecting it will soon be broken by quantum computers. Although no machine today can break elliptic curve cryptography or RSA, quantum hardware is advancing rapidly and will inevitably change how organizations protect their data. Ciphertext and credentials captured by

<pre style='white-space:pre-wrap;width:81ex'>fix #95519: [Bug]: Fallback should trigger on provider upstream_error / LLM request failed (#95542) * fix(agents): fallback on upstream provider errors * refactor(agents): clarify provider fallback classification --------- Co-authored-by: Altay <altay@hey.com></pre>

Google was forced to cap Meta's use of Gemini AI due to a lack of capacity.

<div class="hs-featured-image-wrapper"> <a href="https://blog.hubspot.com/marketing/what-is-ai-search-optimization" title="" class="hs-featured-image-link"> <img src="https://53.fs1.hubspotusercontent-na1.net/hubfs/53/what-is-ai-search-optimization-1-20260627-593826.webp" alt="Two women at laptop researching what is AI search optimization" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"> </a> </div> <p>AI search optimization is the practi

<div class="hs-featured-image-wrapper"> <a href="https://blog.hubspot.com/marketing/scrunch-vs-semrush" title="" class="hs-featured-image-link"> <img src="https://53.fs1.hubspotusercontent-na1.net/hubfs/53/scrunch-vs-semrush-1-20260625-9605726.webp" alt="scrunch vs semrush: woman on laptop researching the two AEO tools" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"> </a> </div> <p>Scrunch vs Semrush comes down to one question: Do you ne

Vercel Functions now support Node.js and Python deployments up to 5GB in package size on Fluid compute , a 20x increase over the previous 250MB limit. Large Functions support is in public beta . This makes Vercel compatible with backend workloads that do not fit within the 250MB limit, such as Python data and AI libraries, large generated clients, browser automation dependencies, image and video processing packages, and routes with substantial shared application code. New projects are automatica

Showcasing the importance of open source innovation in American AI, Palantir’s new intelligent engine — introduced today — uses NVIDIA Nemotron open models to serve the needs of U.S. government agencies. Open source software has long been a pillar of U.S. technology leadership. In 1969, DARPA connected four university computers — from UCLA, Stanford, UCSB […]

We’ve taken one small step towards robot police officers: a drone capable of disarming a suspect: In a June 22 video posted on the Sacramento County Sheriff’s Office’s Instagram page, an officer wearing goggles can be seen operating a drone to retrieve a knife from an armed suspect hiding inside a cluttered house. “After not responding to negotiators, a drone was deployed inside the residence,” the post says. “Drone pilots located the suspect hiding in a corner of a garage” and then used a high-

In production on millions of boxes and the payoff is a 25% reduction in machines needed for some inference workloads

As an outsider to the US ecosystem, a lot of the surrounding startup activity feels strange to me. Many companies seem to be building thin wrappers, agent orchestration layers, AI-generated apps, or B2B tools whose main customers appear to be other AI startups. At the same time, the social and funding environment seems extremely intense. Everyone is excited by “solving a problem,” everyone is racing toward YC or seed funding like it’s the new Stanford acceptance letter. Do people inside this SF

"found" cosmic horror game with malicious RPG mechanics forces player to roll 9 digit RNG to earn $ to move and interact. entire game was written as a single html file by Claude Code on mobile, revised as a codebase on desktop with Codex, then wrapped with electron as proof-of-concept. lore written using Blair Witch Template to beef market ability and implied cognitohazard. project initialize: 4/30/26 project prototype: 5/15/26 project beta test & debug: today market release: 7/17/26 total days

With AI-powered noise reduction, an automatic speech-focusing system, and a simple, effective hearing test, it’s a shame these aids don't come with a better app.

Hi there HN! Been working on agent-swarm since Nov last year, and talking with a lot of companies on the exact same narrative that Claude Tag and Karpathy are pushing as "breaking change" and "the new UX era of AI". We believe that as you move to an AI-native approach in your company you should be in control of the models, harnesses, context, and memory. It should not be a blackbox as Claude is doing. Hence the swarm is trying to solve this problem: and open project that let you define a team of

Article URL: https://seventeenlabs.io Comments URL: https://news.ycombinator.com/item?id=48717293 Points: 1 # Comments: 0

Article URL: https://github.com/bastion-soft/pi-detector-bench Comments URL: https://news.ycombinator.com/item?id=48717269 Points: 2 # Comments: 0

A persistent, file-based memory layer for coding agents — give Claude Code, Codex & others a project brain (durable decisions, requirements, constraints) via a zero-dependency CLI. Comments URL: https://news.ycombinator.com/item?id=48717245 Points: 3 # Comments: 0

PARTNER CONTENT: • Supporting on-premise local AI large model operation, the solution cuts cloud costs and secures sensitive corporate data for SMEs

<!-- SC_OFF --><div class="md"><p>Google deployed an agentic AI peer-reviewer at two top CS conferences — reviewing ~10,000 papers with 30-minute turnaround — and the new formal research paper shows it catches 34% more mathematical errors than zero-shot prompting; the precedent for AI-automated scientific review at conference scale is set and now formally documented.</p> <p>--</p> <p>Source: <a href="https://arxiv.org/abs/2606.28277">https://arxiv.org/abs/2606.28277</a></p> </div><!-- SC_ON -->

PARTNER CONTENT: Showcased at MWC Shanghai 2026, the local AI-driven solution features camera-free whole-home sensing and advanced DDoS protection to unlock new operator revenue and redefine smart home security

Security researchers at Mozilla's 0DIN platform have shown how a single compromised GitHub repo can take over a developer's machine the moment an AI coding tool like Claude Code runs its setup. The catch: the malicious code only loads at runtime via a DNS query, invisible in the repo, to scanners, and to the AI agent itself. The article Claude Code runs a GitHub repo's hidden malware without verification, giving attackers full control appeared first on The Decoder .

We desperately need new options. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

The All England Lawn Tennis Club is adding new AI-powered features to Wimbledon’s digital platforms through its ongoing work with IBM. The updates will be available through the Wimbledon app and wimbledon.com as first-round matches begin on Monday. They include an upgraded Match Chat assistant and a new feature called Key Moments. The features will […] The post Wimbledon adds IBM AI tools for live match coverage appeared first on AI News .

Article URL: https://github.com/ohad6k/VibeRaven Comments URL: https://news.ycombinator.com/item?id=48717113 Points: 6 # Comments: 2

<pre style='white-space:pre-wrap;width:81ex'>fix(codex): preserve per-app approval reviewer policy (#97327) * fix(codex): version app inventory cache keys * fix(codex): refresh and replay plugin app policy * fix(codex): preserve user reviewer for plugin turns * fix(codex): gate plugin reviewer from startup policy * fix(codex): route app approvals to user reviewer * fix(codex): prompt destructive app tools * fix(codex): scope app approval reviewers * test(codex): complete app policy fixture * fix

<p>🚀 After months of development, I'm excited to launch InterviewPro.</p> <p>InterviewPro is an AI-powered interview preparation platform designed to help students and professionals prepare for technical interviews more effectively.</p> <p>It includes:</p> <p>✅ AI Mock Interviews</p> <p>✅ Company-specific interview questions</p> <p>✅ Resume Review and ATS Analysis</p> <p>✅ System Design preparation</p> <p>✅ DSA and coding challenges</p> <p>✅ Learning roadmaps for multiple technologies</p> <p>My

Article URL: https://github.com/matthiasnordwig/pdf-struct-chunker Comments URL: https://news.ycombinator.com/item?id=48716707 Points: 2 # Comments: 0

<h2> How visual Python learning can make algorithms feel less scary </h2> <p>I’ve always felt that algorithms become harder than they need to be when they are explained only in text.</p> <p>A student can read the same sorting logic a few times and still not really <em>see</em> what is happening. The idea sounds simple on paper, but once variables start changing and loops begin nesting, things can get confusing very quickly.</p> <p>That is one reason I started thinking more seriously about visual

<h2> The Problem You Probably Have Already </h2> <p>If you're building agentic systems in 2025, chances are you've already got two integration layers in your stack:</p> <ul> <li> <strong>MCP</strong> (Model Context Protocol) — wiring your AI models to databases, APIs, filesystems, internal tools</li> <li> <strong>A2A</strong> (Agent-to-Agent) protocols — letting your agents discover, negotiate with, and invoke <em>each other</em> </li> </ul> <p>They're not competing. They're complementary. But w

<h2> The Pipeline That Worked Too Well </h2> <p>I had to kill a pipeline that was doing exactly what it was supposed to do. It was rewriting job descriptions at scale, improving SEO, and running without errors. The client asked me to shut it down anyway.</p> <p>The problem wasn't quality. It was cost. Running GPT-class models across a million listings added up faster than anyone expected. The pipeline worked perfectly, and that was the problem. Every perfect run cost money. Over time, the bill b

<table> <tr><td> <a href="https://www.reddit.com/r/MachineLearning/comments/1uin5ad/i_made_a_quiz_that_tells_you_which_llm_you_align/"> <img src="https://preview.redd.it/yx86ia6rr6ah1.png?width=140&height=80&auto=webp&s=50a7c238e71f794f9908533538785f72e88913a9" alt="I made a quiz that tells you which LLM you align with most, based on personality and values research across 15 models [R]" title="I made a quiz that tells you which LLM you align with most, based on personality and values research ac

Natural language processing is reshaping professional communication on online platforms, enabling more relevant and personalised networking interactions. As AI-driven systems increasingly comprehend and generate human language, these technological advances affect how users pursue and maintain professional connections, presenting both opportunities and challenges in authentic relationship-building. Professional networking now relies on a growing ecosystem of AI-powered […] The post Advances in Na

As of June 29, the Roborock Qrevo Slim robot vacuum and mop is on sale at Amazon for $699.99. That's 50% off its list price of $1,399.99.

Modern DevSecOps needs security checks that run before release day. Teams now write code, build services and deploy updates at a pace that manual review cannot match. That’s why they use automated testing, as it helps catch routine flaws before they reach production. The pressure has grown. Verizon’s 2025 Data Breach Investigations Report found that […] The post Best Automated Security Testing Tools for Modern DevSecOps appeared first on AI News .

Article URL: https://api.octoperf.com/doc/mcp/ Comments URL: https://news.ycombinator.com/item?id=48716592 Points: 1 # Comments: 0