Software Engineering News
Pi 1.0, a minimal agent harness, is now generally available, alongside an experimental package called Pi Durable for building long-running agentic applications, from hackernews.
Pi 1.0 is the stable release of the open-source agent harness that has been in wide use for months. New features include native support for MCP and non-LLM models (such as image models and Jev), virtual model extensions, deferred tool loading, cache warming for Anthropic models, and mid-conversation system message changes. A new TUI theme and full-screen mode default are also in. The project’s maintainers frame the release as the result of a slow adoption process — features had to stick before being added.
The same team shipped Pi Durable, an experimental substrate for longer-lived agentic tasks. It is a separate package, not a version of Pi 1.0, and is aimed at builders who need to steer intelligent agents outside the terminal and across sustained conversations. Both packages are MIT licensed; docs live at pi.dev and code at github.com/earendil-works/pi
Source: news.ycombinator.com, earendil.com
A review of the Shimano Bicycle Museum in Sakai, Japan, praises its focus on bicycle history over corporate promotion, from hackernews.
The museum, housed in a modern glass-and-steel building near Osaka, walks visitors through bicycle evolution—from hobby horses to modern race bikes—in a darkened hall with spotlit displays. It’s bilingual, and the tour flows into a bright room where bikes sit on low stages, not behind glass, letting you inspect details like Freddy Maerten’s steel Flandria with its Shimano Crane derailleur, a Paralympic bike with custom modifications, and a Specialized Stumpjumper built from early mountain bike parts. The collection also includes everyday machines: a Swedish postal bike with studded tires, a dust-caked Cannondale from a 55,000km world tour, and a scratched Merida from the 2021 Olympic road race.
Shimano’s own history is sparse—only a brief mention of gun blacksmiths in Sakai repairing imported bikes in the late 19th century, and a small display of the company’s first product, a freewheel. The museum does showcase Shimano’s Dura-Ace and XTR groupsets from 1973 onward, with empty cases reserved for future models, plus a stand that explodes a race bike into its 3,200 parts. A library and locked archive hold catalogues and brochures for researchers, and there’s no gift shop push.
The verdict calls it an enjoyable, peaceful celebration of the bicycle, where the machine comes first and corporate hard sell is absent. Admission is around JPY 500 (under €3), though the reviewer suspects it may have gone up slightly. Sakai is a 30-minute train ride from Osaka’s central station, with bike parking on site.
Source: news.ycombinator.com, inrng.com
Debian released a security update fixing hundreds of vulnerabilities in the Linux kernel, from hackernews
Debian published DSA-6528-1 on September 29, 2026, addressing a massive batch of security issues in the `linux` package. The advisory lists more than 400 CVEs, including older entries like CVE-2024-52560 and a wave of newly assigned IDs from late 2026. The update applies across Debian's supported releases, with maintainer Salvatore Bonaccorso signing the announcement.
Debian's advisory does not detail what each of the listed CVEs covers. The sheer volume suggests the update folds in many subsystems—drivers, networking, filesystems, core memory management—rather than a single critical flaw. People running Debian systems iside the stable or LTS branches should treat the `linux` package upgradea mandatory maintenance task. Checking the full CVE list on Debian's security tracker remains rotating pointers as to whichf exposure matters most in production contexts. The patchcompression also points toward the kernels hardening posture holding upto newly reported abuse paths, though specific exploitation severity isn't made public here.
Source: news.ycombinator.com, lwn.net
Cloudflare released open-weight decision models Clef and Clef-flash, plus a new RL fine-tuning platform, from hackernews.
Cloudflare launched two decision models, Clef and Clef-flash, hosted on Workers AI and open-sourced under Apache 2.0 on Hugging Face. Decision models return bounded, typed outputs with probabilities for routing or escalation tasks, distinct from open-ended LLMs. Clef is Jev-API compatible and leads the Jev Decision Index on several benchmarks, including a 98.47% exact-match score on BFCL and 71.69% nDCG@10 on ToolRet. The models include a vision encoder, a 64k context window (versus Jev’s 32k), and median latency of 209ms for Clef and 38.8ms for Clef-flash.
Clef beat Jev on 3 of 4 categories in Typesafe’s eval suite, including invoice processing (64.7 vs. 61.8) and customer service (76.3 vs. 76.0), though Jev led on When2Call accuracy (80.97% vs. 72.37%). In a Cloudflare test, Clef classified website domains in 2.2 seconds including fetch and render, versus 4.7 seconds for gpt-oss-120b, and returned more categories. The company positions the models for agentic workflows like ticket routing or threat intelligence, with an enterprise guarantee that requests aren’t stored or trained on.
Cloudflare also debuted a reinforcement learning fine-tuning product. It’s initially available through a forward-deployed engineering team for hands-on tuning, with a self-serve platform planned later so customers can train and redeploy Clef onto Cloudflare’s edge GPUs. The training approach adapts DiffusionGemma to output deterministic probabilities via logprobs, as described in a demo Cloudflare posted the same week Jev launched.
Source: news.ycombinator.com, blog.cloudflare.com
Singapore's GovTech launched FirstDate, a pilot government-run dating service for civil servants using the Nobel-winning Gale-Shapley algorithm, as reported from hackernews.
GovTech's FirstDate pilot is live for government workers, three years after the Social Development Network was scrapped. Participants fill a 30-question survey across eight categories, then get matched in cycles via the Gale-Shapley Stable Marriage Algorithm, which ranks opposite-gender participants by questionnaire compatibility. Singpass verifies identity, and contact details swap only if both sides agreeto the match.
.
The questionnaire covers hobbies, diet, travel habits, sleep schedules, smoking, religion, love languages, and hard nos, plus a short note for the future match. A earlier prototype proposed free meals for first dates, which likely explains the food-heavy questions. The algorithm guarantees stability — no two people would both prefer each other over their assigned partners — but it aims to prevent preference-switching rather than promise happiness.
The pilot's data use remains unspecified beyond the matchmaking itself, though the algorithm's history in medical residency matching suggests a preference-ranking logic underneath the survey answers. Three structural limits stand out: stability is not satisfaction; the algorithm demands two equal-sized groups, which may not reflect real dating pools; and the cycle-based format caps matches per round, stranding participants until the next programmatic round.
Source: news.ycombinator.com, singapore-samizdat.com
SvelteKit 3.0 is out with config moved to vite.config.ts, a new #lib alias, and improved error handling, from hackernews.
The Svelte team shipped SvelteKit 3.0, the official framework for Svelte, with an emphasis on polish and type safety. Major breaking changes include configuration moving from svelte.config.js to vite.config.ts, the $lib alias becoming #lib to use standard subpath imports, and stronger environment variable handling. Service workers need less boilerplate, and error handling is improved across routing and data loading.
Migration is handled by the `sv migrate sveltekit-3` command, which auto-fixes most codebases and spits out a TODO list for the rest. New apps start with `sv create`. Remote functions—the team’s stated top priority—are not here yet; they require the Async Svelte experimental flag, so the team is asking for patience while they finish that piece. In-person Svelte Summit happens November 19-20 in Ljubljana, Slovenia, also celebrating Svelte’s 10th birthday.
Source: news.ycombinator.com, svelte.dev
Earendil released Pi 1.0 and a new experimental package called Pi Durable, a durable harness for long-running agents, from hackernews.
Pi 1.0 ships after years of hardening, aimed at the existing Pi coding agent that runs in a terminal, driven by one person. The release pairs with Pi Durable, a separate framework for building any agentic application, coding agents included, built to run anywhere there is a JavaScript runtime and survive crashes, restarts, and multiple human steerers.
Pi Durable is designed for infinitely long conversations, with storage backends for memory, SQLite, JSONL, and accommodation for custom backends like Postgres or key-value stores. Tools execute in environments that can be local or remote, decoupling the harness from execution.
The framework's storage code avoids Node APIs, so it runs on Bun or inside Cloudflare Durable Objects. It checkpoints every step of a run, resuming unfinished tasks after process death —model requests are resent, partial answers marked aborted, and safe tool calls rerun. Subagents are easy to implement as separate conversations, continuing from where they left off. The entire source, about 15,000 lines, is designed to be understandable to the agents themselves, with foundational storage at 3,000 lines usually skippable.
Source: news.ycombinator.com, earendil.com
Remocn/remocn-studio is trending in typescript (75 stars) from GitHub
Agent-native video editor
Source: github.com
bangtutorial/bang-story is trending in typescript (25 stars) from GitHub
Aplikasi desktop untuk mengubah ide cerita jadi video YouTube dengan AI. Masukin ide cerita: naskha, gambar, video jadi otomatis.
Source: github.com
henrydennis/petal is trending in rust (43 stars) from GitHub
A fast, exact disk-space explorer for macOS with a live sunburst chart. Rust + GPUI.
Source: github.com
mkdir700/wx-ime-sdk is trending in rust (37 stars) from GitHub
微信输入法（WeType）跨设备剪贴板的独立 Rust 客户端
Source: github.com
BootLoops-ai/bootloops is trending in python (61 stars) from GitHub
BootLoops 1.0: certified computational tools and house engines for exact and high-precision physics and quantitative science, built to be driven by LLM agents. MIT; docs CC BY 4.0.
Source: github.com
ACoci86/terrahour is trending in python (60 stars) from GitHub
A world clock for the terminal, with a day and night map and a 24-hour timeline to find a time that works across time zones.
Source: github.com
XavyProd/pkg-mutant-shop is trending in python (49 stars) from GitHub
PKG MUTANT SHOP - a homebrew PKG store and installer for jailbroken PS5 and PS4, with a Windows companion.
Source: github.com
AI News
OpenAI argues that the real economic impact of advanced AI will come from automating the unglamorous execution work behind breakthroughs, not from the ideas themselves, from OpenAI Blog.
The post frames "the eternal complement" as the dynamic where AI handles the routine legwork—data prep, iteration, testing, logistics—that currently bottlenecks turning a novel insight into a working product. That execution layer, not the flash of inspiration, is where most productivity is lost and where AI-driven speedups could compound.
atically.
The piece ties this to the broader economy: if the cost of ordinary execution falls toward zero, then thepace of progress bends upward—new drugs, materials, or software reach market years earlier. OpenAI pitches this not as replacement of human judgment, but as itsenablement: people keep the strategic questions; machinesabsorb the tedious middle. What happened in the article is a forward-looking argument, not a product launch or a policy shift—but the position fills out the company’s stated bet on agentic, task-level AI over chat-onlytools. The post closes on a pragmatic note: the spoils will go to whoever can harness AI for the bulk of unglamorous work, not just for the occasional eureka.
Source: openai.com
Albertsons Cos. is deploying ChatGPT Enterprise and the OpenAI API to speed up internal work and improve the grocery shopping experience for millions of customers from OpenAI Blog.
Albertsons is rolling out OpenAI’s ChatGPT Enterprise and API toolsacross its operations. The company expects the technology to cut down on time spent on routine tasks, freeing up staff to focus on customer-facing work. It’s a straight adoption play: get the tools into employees’ hands and let them find the efficiencies.
— not a pilot or research project, but an enterprise-wide integration aimed at day-to-day retail grind.
.
On the customer side, the API is being folded into digital shopping surfaces. That means smarter search, more relevant product discovery, and—presumably—less friction when someone’s trying to build a weekly meal plan or track down an ingredient at 6 PM. Grocers have tight margins and thin digital patience, so the payoff hinges on whether these features actually reduce drop-off, not just look neat in a demo.
.
The move puts Albertsons alongside a growing list of non-tech companies treating generative AI as core infrastructure, not an experiment. Their bet is that internal speed boosts and external personalization will compound into real loyalty—though neither effect has been publicly measured yet.
Source: openai.com
The Den social club cut grant application time from 3 days to 2 hours and liquor-license work from ️4 days to ️3 hours using ChatGPT Work, per from OpenAI Blog.
The Den, a social club opening a new location, has been using ChatGPT Work to shrink administrative workloads. Grant applications that once took three days now close in two hours; liquor-license materials dropped from four days to three hours.
The club’s leadership credits the freed time—roughly 10-15 hours a week—to growth work like scouting new sites and preparing the launch instead of grinding through paperwork. Their new location is in progress as the tooling handles compliance drafting and submission prep.
.
ChatGPT Work is OpenAI’s enterprise-tier offering, aimed at organizations that want AI inside their workflows with more control over data and permissions. The Den’s case is an early, concrete example of how that tier plays out beyond corporate departments—here, a brick-and-mortar social venue using it to move faster on regulatory and funding tasks.
Source: openai.com
OpenAI says it disrupted a coordinated campaign aimed at extracting protected model reasoning and is now hardening defenses against such distillation attacks from OpenAI Blog
A coordinated attempt to extract reasoning from one of OpenAI’s models was detected and thwarted. The campaign appears to have used adversarial techniques—likely building a custom pipeline to query the model in ways intended to surface hidden chain-of-thought or other protected internal logic.
and then distill that output into outer model. OpenAI did not disclose which model was targeted, nor the scale of the operation, but characterized the campaign as deliberate and coordinated rather than opportunistic.
.
In response, OpenAI says it is strengthening its defenses against adversarial distillation—a class of attacks where someone uses a model’s outputs to train a copy that bypasses protections on reasoning or other sensitive behavior. The mitigations include both improved detection and harder guardrails around how reasoning is exposed under unusual query patterns. The company did not share technical specifics, keeping details close to avoid aiding future attempts, but framed the work as ongoing rather than a one-time fix.
.
The episode is partly notable for being flagged publicly at all: distillation attacks von models are not nuevos, but most go unnamed. By naming this one, OpenAI signals it now treats coordinated reasoning-extraction as a first-class threat—one worth disrupting in real time and shoring up against, rather than cleaning up aftercharge.
Source: openai.com
Hugging Face Blog introduces AutoSynthData, a method for automatically generating training data tailored to enterprise AI agents, with details inferred from the title.
The post centers on AutoSynthData, a framework aimed at producing synthetic datasets for enterprise-grade agentic systems. Its core premise is that hand-labeled data doesn't scale for the specialized, multi-step tasks these agents handle, so generation becomes the bottleneck.
. The title implies a pipeline that turns raw requirements or task descriptions into structured, usable training examples, likely via LLM-driven synthesis loops with validation checkpoints.
.
That pipeline presumably cuts the cost and time of manual curation, letting teams spin up task-specific corpora without waiting on human annotators. For enterprise deployments, this points toward faster iteration on agent behavior — retraining on fresh, domain-relevant scenarios as workflows shift, rather than rebuilding data assets from scratch. The exact mechanics — whether it uses seed prompts, retrieval-augmented generation, or quality filters — aren't detailed in the available text, but the focus is clearly on operationalizing synthetic data at scale.
Source: huggingface.co
Hugging Face announced Olmo-core 3, an open and scalable training infrastructure designed for large mixture-of-experts models, from Hugging Face Blog.
The release centers on infrastructure rather than a finished model. Olmo-core 3 is positioned as a training stack built to support large-scale MoE architectures, with openness as a core constraint — the goal is to let others run similar experiments without proprietary tooling.
.
The project emphasizes scalability of compute and data pipelines, targeting the efficiency gains MoEs offer when trained at scale. By publishing the infrastructure, Hugging Face aims to lower barriers for research groups wanting to push past dense-model limits.
.
No benchmark results or model weights are attached to this announcement; the concrete artifact is the tooling itself. The framing suggests follow-up releases may include actual MoE checkpoints trained on this stack, but that remains ahead rather than today's news.
Source: huggingface.co
Hugging Face introduced a new Open TTS Leaderboard for scalable, standardized evaluation of multilingual text-to-speech and voice cloning models, from Hugging Face Blog.
The leaderboard provides a public, reproducible benchmark for comparing open-source TTS systems across multiple languages. It uses automated metrics rather than relying solely on human listening tests, enabling faster iteration and broader model coverage. The framework also targets voice cloning performance separately, tracking how well models preserve speaker identity alongside speech quality.
The initiative addresses fragmentation in the TTS evaluation space, where researchers previously had to juggle inconsistent datasets and scoring methods. By standardizing tasks, the leaderboard makes results directly comparable and lowers the barrier for new open models to demonstrate progress. The blog points to the companion code and datasets as the mechanism for running evaluations consistently, though no numerical results or model rankings are included in the available article text. The announcement positions this as a practical infrastructure step for the field, not a one-time competition
Source: huggingface.co
NVIDIA released Kumo Tabular, a model that sets a new accuracy-efficiency benchmark for tabular prediction, from Hugging Face Blog
NVIDIA has introduced Kumo Tabular, a tabular prediction model that claims to outperform existing approaches on both accuracy and computational efficiency. The model was announced via the Hugging Face Blog, though theg technical specifics weren’t included in the available text.
.
DETAIL Kumo Tabular targets the persistent problem of tabular data — the workhorse format of most enterprise analytics — where tree-based ensembles have long dominated. NVIDIA’s entry signalsan effort to push neural approaches past those baselines without sacrificing speed or scaling costs.
The announcement comes as part of NVIDIA’s broader push into applied AI infrastructure. Kumo Tabular’s positioning suggests the company is betting on specialized models for structured data, rather than only LLMs or vision systems. That lanes remains crowded with tools like XGBoost and TabNet, so Kumo will need to demonstrate clear wins in real workloads to gain traction.
No benchmark numbers, architectural details, or release dates were provided in the article text, leaving the precise nature of “accuracy-efficiency frontier” open to interpretation. What is clear is NVIDIA sees tabular modeling as a strategic gap worth filling — one where winning could mean default choice for thousands of ordinary data teams.
Source: huggingface.co
Trump reportedly asked Musk’s AI chatbot Grok for advice before moving to invade Venezuela and capture Nicolás Maduro, according to TechCrunch AI.
The report centers on a single claim: that President Trump consulted Grok prior to taking military action against Venezuela’s government. No further details were provided in the article about what Grok recommended, or whether its output influenced the decision. The source material is thin—essentially one sentence with no corroborating quotes or official confirmation.
The claim surfaces amid broader scrutiny of Musk’s access to the president and the role of Grok in high-stakes decision-making. TechCrunch’s framing positions the anecdote as evidence of AI’s growing presence in executive processes, though the article stops short of verifying the exchange independently.
No response from the White House, Musk, or Grok’s development team is included. The omission leaves the matter as an unconfirmed report, pending further reporting or denial.
Source: techcrunch.com
OpenAI is adding shopping features to ChatGPT that let users virtually try on clothes and save items to a favorites list, from TechCrunch AI.
The new features let ChatGPT users upload their own photos to see how clothing and accessories might look on them, directly within the chat interface. Products that catch a user’s eye can be saved to a Favorites library, effectively turning the chatbot into a personal shopping assistant.
This is a practical expansion of ChatGPT’s utility beyond text generation, moving into e-commerce territory where visual and personalized recommendations matter. OpenAI hasn't announced a partner list or a timeline for broader rollout, but the capability is live now for some users.
The shopping tools arrive as OpenAI accelerates feature releases across consumer and enterprise surfaces. For retailers, it signals a shift toward conversational, image-driven discovery — but for now, the experience is contained inside ChatGPT’s existing app and web product.
Source: techcrunch.com
Google launched its first advanced chip into orbit to explore space-based data centers, a step it says depends on SpaceX’s Starship flying roughly 1,800 times to become practical, from TechCrunch AI
Google sent an AI accelerator chip to orbit on an early SpaceX rideshare mission, testing whether commercial silicon can survive radiation and other space conditions. The flight is a proof-of-concept, not a working data center — theardware stayed aloft briefly, with telemetry returning to Google engineers on the ground.
.The company framed the mission as a low-cost experiment to gauge whether off-planet computing could eventually offload ground data centers.
.
Space data centers would need massive launch capacity—Google’s own analysis pins that at around 1,800 Starship flights—to haul up enough servers and power infrastructure for any meaningful workload. That figure dwarfs current orbital launch rates by orders of magnitude, making the near-term economics speculative at best rather than a looming shift in cloud infrastructure. For now, Google’s bet is narrower: validate that chips can fly, then wait for Starship’s reuse curve to bend.
Source: techcrunch.com
OpenAI has dismissed three safety researchers after an internal investigation found they mishandled sensitive company information, according to a WSJ report from TechCrunch AI.
The three researchers were let go following a probe into how they handled confidential material. OpenAI’s review concluded the individuals did not meet the company’s internal standards for protecting sensitive data.
, the report said.
.
The departures come amid broader scrutiny of OpenAI’s safety practices and its push to tighten control over internal information. The company has not publicly detailed freda specifics of the alleged mishandling, nor freda names of freda researchers involved, per freda WSJ reporting.
Source: techcrunch.com
Suno launched a public beta of its new Speech feature, which generates spoken-voice audio from scripts or descriptions and can layer it over AI music, from The Verge AI.
Suno, best known for its AI music generation, is adding spoken-word audio to its toolbox. The new Speech feature is now in public beta across Suno’s web and mobile apps, letting users generate voiceovers from text prompts or script uploads. It can also combine those voices with background music in one generation step.
.
Product chief Jack Brody framed the expansion as a natural broadening of the company’s mission, saying music remains the core but that Suno’s vision “has always extended to other forms of human expression.” The feature represents a shift from pure melody toward more mixed audio formats, positioning Suno against dedicated voice-cloning and text-to-speech tools while keeping its existing music-generation engine as the foundation.
Source: theverge.com
Google launched Guided Vision in Gemini Live on Android today, using the camera to provide real-time audio descriptions for reading, identifying objects, and navigating surroundings, from The Verge AI.
Guided Vision is now rolling out in Gemini Live on compatible Android devices. Users can share their camera feed within the app, and Gemini will narrate what it sees — covering small text, object identification, spatial descriptions, and details on specific items. The feature is positioned as an aid for blind or low-vision users, though it also targets situational help like reading fine print.
The launch parallels Apple’s VoiceOver Live Recognition on iPhone and Vision Pro, but Google’s version lives inside the existing Gemini Live interface rather than a dedicated accessibility tool. Distribution is via the Gemini app on Android; the Verge’s report notes availability is tied to compatible devices, with no mention of iOS or broader platform support at this time.
Source: theverge.com
A federal judge dismissed antitrust lawsuits from Chegg and Penske Media Corporation accusing Google of coercing publishers into feeding its AI Overviews for free, siding with Google on Wednesday from The Verge AI.
US District Judge Amit Mehta ruled that the claims from Chegg and Penske Media Corporation —parent of Rolling Stone—don't hold up under antitrust law. The publishers had filed suit last year, alleging Google abused monopoly power by forcing outlets to supply content for AI Overviews without compensation, with the implicit threat of vanishing from search results.
.
Mehta's decision lands on Google's side, shutting down the pair of cases before they advanced further.. The ruling suggests the alleged conduct, however commercially fraught for publishers, didn't amount to an antitrust violation as argued..
Chegg, an education company, and Penske Media had framed Google's practices as coercive—effectively demanding free labor from publishers under duress of losing search visibility.. The judge disagreed, finding the claims didn't satisfy the legal standards required to proceed..
Source: theverge.com
OpenAI unveiled Dots, a GPT-6 Astra-powered AI agent, at DevDay 2026 as a direct challenge to Meta's Muse platform, from The Verge AI.
OpenAI used its DevDay stage to position Dots as a "real-deal AI" agent, contrasting it with Meta's rapidly adopted Muse platform. CEO Sam Altman leaned into movie-inspired aesthetics, showing Dots as colorful, customizable blobs with eyes — a deliberately cute counter to Muse's similar design language. The demo went beyond simple assistant tasks, showing Dots generating metaverse-style virtual environments, an area where OpenAI has previously lagged.
ne key weakness surfaced: the article notes Dots "strikingly" falls short in one unspecified area—ethe free tier. Meta's Muse has surged on its no-cost availability, while OpenAI's pricing model for Dots remans unclear, hinting at a potential competitive disadvantage before broad rollout. OpenAI's broader strategy appears calibrated tob check Meta's momentum in consumer AI agents, but the cost question hangs over whether cutenessand worldbuilding trumps free access.
Source: theverge.com
DeepSeek released an open-source desktop harness app for macOS andWindows in public preview, built on a plugin architecture, from hackernews.
DeepSeek Harness is now available as a public preview worldwide and open source, running as a desktop app for macOS andWindows or from a web UI launched via npx. It's built on Cordis's “everything is a plugin” architecture, letting users extend tools, skills, and the interface with plugins—including creating them through chat in “Creator mode.”
The harness covers everyday work like organizing files, analyzing data, drafting docs, and creating slides, plus coding tasks such as exploring repos, fixing bugs, building features, and running tests. Research and background tasks—finding info, verifying facts, citing sources, running scripts, batch-processing files—are also included, along with a scheduled tasks plugin for recurring work.
开发者工具 let users inspect execution traces and detailed runtime information to troubleshoot tool calls. Setup starts with one command—install Node.js, then `npx @deepseek-ai/dsh web`—or clone the full source from GitHub at deepseek-ai/deepseek-harness.
Source: news.ycombinator.com, deepseek.com
Politics
Coverage by lean: ● covered it · ○ not in this digest's sources
| Story | Left | Center | Right |
|---|
| Tennessee execution attempt fails after lethal injections | ○ | ● | ○ |
| A 'dark cloud' and 'dismal' mood: How it feels on Cornell campus right now | ○ | ● | ○ |
Are Democrats Worried About the Wrong Rich People? from NYT
As billionaires become a popular political target, a new book urges Washington not to lose sight of a much larger, and quieter, group of wealthy business owners.
Source: NYT (lean-left, nytimes.com)
AIPAC Is Spending Millions to Elect Friends. It’s Also Making Enemies. from NYT
The American Israel Public Affairs Committee has increasingly become a campaign issue unto itself. Some candidates have worn opposition from AIPAC as a badge of honor or made a point of rejecting the group’s support.
Source: NYT (lean-left, nytimes.com)
Trump Puts on a Midterms Show for an Audience That’s Already Sold from NYT
President Trump spent the day preaching to his most fervent fans, not the voters he will need to help elect or re-elect a slate of Republicans fighting in states from Alaska to Maine.
Source: NYT (lean-left, nytimes.com)
Military Leaders Will Examine Ways to Limit Civilian Deaths in War from NYT
U.S. Central Command will hold a daylong conference at its Tampa headquarters after seven months of war in Iran.
Source: NYT (lean-left, nytimes.com)
Wind Warnings Were Never Given to Pilots in Miami Crash, Federal Report Says from NYT
The pilots of an Amazon cargo jet had never worked together before the day of the crash, and were traveling at an abnormally high speed when they hit the runway, the report said.
Source: NYT (lean-left, nytimes.com)
Trump vows Iran war to end "very quickly" as more troops head to Mideast from CBS News
Some 9,000 more U.S. forces head for the Middle East as Trump vows the war will end "right after the election," if not sooner.
Source: CBS News (lean-left, cbsnews.com)
Trump likely to pick Jay Clayton for AI czar, sources say from CBS News
Jay Clayton will likely be the White House's pick for AI czar, and the Trump administration has been discussing having him remain in his current role as director of national intelligence, according to sources briefed on the matter.
Source: CBS News (lean-left, cbsnews.com)
Homan defends ICE crackdown: "Give me an example of where we went too far" from CBS News
As the Trump administration's crackdown on illegal immigration draws pushback from some GOP lawmakers, White House border czar Tom Homan told CBS News critics should "educate themselves."
Source: CBS News (lean-left, cbsnews.com)
Trump promotes tariffs, tries to shore up support for Ken Paxton in Texas from CBS News
President Trump has launched a campaign blitz for Republicans in the run-up to the midterm elections, with some unexpected early stops in deep-red Oklahoma and Texas. Nancy Cordes reports from Denton, Texas.
Source: CBS News (lean-left, cbsnews.com)
Judge blocks Trump's firing of top federal prosecutor in Seattle from CBS News
Attorney General Todd Blanche announced in July that President Trump had fired Roger Rogoff as the U.S. attorney in Seattle, less than an hour after he was sworn into the post.
Source: CBS News (lean-left, cbsnews.com)
Can the Democrats take Texas? – podcast from The Guardian
Democratic Senate candidate James Talarico is surprising everyone, as polls have him leading his Republican opponent, Ken Paxton, in the Republican-led state of Texas.
Jonathan Freedland speaks to James Henson, founder and executive director of the Texas Politics Project, about why a blue wave in the Lone Star State would be seismic
Watch Jonathan Freedland on our new Politics Weekly YouTube channel
Send your questions and feedback to politicsweeklyamerica@theguardian.com
Watch Stateside’s interview with Mahmoud Khalil and his wife, Noor Abdalla
Continue reading...
Source: The Guardian (lean-left, theguardian.com)
The end times fascists are plotting their escape from our world. Our mission is clear – podcast from The Guardian
An unholy alliance is working toward an apocalyptic future from which only a chosen few will emerge unscathed. It’s becoming clear how we stop them
By Naomi Klein and Astra Taylor
Read the text version here
Support the Guardian today: theguardian.com/longreadpod
Continue reading...
Source: The Guardian (lean-left, theguardian.com)
Trump doesn’t rule out sending ICE to polling sites; US supreme court to review policy to detain immigrants challenging deportation – as it happened from The Guardian
This blog is now closed.
Sign up for the Breaking News US email
The White House has, again, blocked CNN from covering Donald Trump as part of the rotating press pool.
The outlet was scheduled to be a part of the group of media traveling to Texas and Oklahoma with the president today, but was removed from list of newsrooms in place for the trip, per the White House schedule.
Continue reading...
Source: The Guardian (lean-left, theguardian.com)
California outlaws child marriage, joining 17 other US states from The Guardian
Advocacy group that pushed for the legislation says 9,000 children are married in California every year
California has outlawed child marriage, joining 17 other US states where the taboo practice is also banned.
Under past state law, minors of any age could obtain a court order for marriage with the consent of their parents.
Continue reading...
Source: The Guardian (lean-left, theguardian.com)
New Washington DC law bans landlords from imposing dog breed restrictions from The Guardian
Landlords can no longer refuse pit bulls, rottweilers, dobermans or other breeds considered ‘scary’
Landlords in Washington DC will no longer be able to refuse a tenant’s dog based on its breed, size or weight, nor can they impose special charges related to these factors.
At least, that’s what a new law that took effect on Thursday aims to achieve. Animal welfare organizations and animal shelters applauded the legislation, emphasizing the benefits of reducing barriers to pet ownership in the US capital.
Continue reading...
Source: The Guardian (lean-left, theguardian.com)
San Francisco's car crime has plunged. How much credit does Flock deserve? from NPR
San Francisco leaders say AI-powered cameras helped drive a dramatic decline in car crime. We examine the evidence — and whether the city can keep its high-tech dragnet without sacrificing civil liberties.
Source: NPR (center, npr.org)
Tennessee execution attempt fails after lethal injections
Center · NPR — It highlighted the botched attempt as unprecedented, blamed state secrecy, and quoted abolition advocates amid renewed national debate. Left out: It did not mention that Governor Lee ordered a third-party review or that the state Department of Correction defended its protocol.
Center · BBC News — It focused on the procedural failure, emergency legal efforts, and victim's family reaction, with detailed medical expert analysis. Left out: It omitted broader national debate statistics or recent legislative efforts to abolish the death penalty.
Source: NPR (center, npr.org), BBC News (center, bbc.co.uk), BBC News (center, bbc.co.uk)
The U.S. Senate is up for grabs. Can this Ohio Democrat win back working class votes? from NPR
After 32 years in Congress, Ohio's Sherrod Brown lost his Senate seat in 2024 as many working class voters turned out for Trump. Now Brown is counting on union voters to send him back to Washington.
Source: NPR (center, npr.org)
A judge permanently dismisses former Olympian's Reflecting Pool charges from NPR
Citing "exceptional circumstances," a D.C. judge has ruled that the federal government cannot refile its Reflecting Pool charges against David Hearn, despite pressure from President Trump to do so.
Source: NPR (center, npr.org)
AI data centers: How much are ratepayers on the hook for? from NPR
Data centers consume large amounts of power and require upgrading infrastructure. But it's difficult to determine how much residential customers benefit and how much they actually pay for the benefit.
Source: NPR (center, npr.org)
A 'dark cloud' and 'dismal' mood: How it feels on Cornell campus right now
Center · BBC News — A 'dark cloud' and 'dismal' mood: How it feels on Cornell campus right now: Students tell the BBC that the tension on campus has become palpable since allegations from 2024 resurfaced.
Center · The Hill — Cornell under a microscope over handling of frat house rape investigation: Cornell University is facing angry calls from politicians, activists and the general public demanding more transparency over how it handled the investigation into a former student's alleged gang rape. Cornell is under a…
The side-by-side comparison wasn't available for this story; each outlet's own headline is shown instead.
Source: BBC News (center, bbc.co.uk), BBC News (center, bbc.co.uk), The Hill (center, thehill.com)
Trump says 'some' munitions are low as Iran war grinds on from BBC News
He also said in an interview with Time magazine that it was possible the US could ramp up strikes on Iran after November's elections.
Source: BBC News (center, bbc.co.uk)
Trump’s powerful new friends are making MAGA uncomfortable from The Hill
President Trump’s appearances with Chinese President Xi Jinping, New York City Mayor Zohran Mamdani (D) and leaders from Silicon Valley risk turning off his political base just as Republicans are seeking to turbocharge turnout ahead of the midterms, say GOP strategists. The president has always loved meeting with the powerful, even if they are dictators...
Source: The Hill (center, thehill.com)
GOP senators warn Trump is hurting Republicans with independents from The Hill
Senate Republicans say President Trump has become increasingly toxic with independent voters critical to winning Senate races in battlegrounds such as North Carolina, Ohio, Iowa and Alaska. Some GOP senators are warning Trump to stay away from key Senate races, cautioning that his presence would likely turn off independent voters. But vulnerable incumbents such as...
Source: The Hill (center, thehill.com)
Younger candidates are finding it tough to break into Congress from The Hill
More than 100 Generation Z candidates launched bids this cycle to run for Congress, per Future Caucus, but few have been successful. The attempts to run highlight how young adults are frustrated with partisan gridlock and want to enter the political sphere, even if their attempts are largely falling short. One possible reason for the...
Source: The Hill (center, thehill.com)
Khanna eyes opportunity in Ocasio-Cortez plotting from The Hill
As Democrats begin to position themselves for the 2028 presidential campaign, few contenders are being watched and analyzed as closely as Rep. Alexandria Ocasio-Cortez (N.Y.), whose decision to enter the race could alter the entire landscape of the Democratic primary. No one has more reason to observe Ocasio-Cortez's every move than Rep. Ro Khanna (D-Calif.), who is frequently...
Source: The Hill (center, thehill.com)
Inside the House GOP's 4-week battle plan to save its majority from Fox News
House GOP Chairwoman Lisa McClain reveals the four key topics Republicans plan to hammer home in the final weeks before the November midterms.
Source: Fox News (lean-right, foxnews.com)
Trump’s first-year government overhaul delivers change Washington hasn’t seen since the 1960s from Fox News
IRS audit revenue dropped 35% after nearly 10,000 agents were let go, raising questions about whether federal workforce cuts are saving money.
Source: Fox News (lean-right, foxnews.com)
Iraq sounds alarm on terror threat from Syria one day after US troops leave from Fox News
Iraq's foreign minister warns ISIS fighters could cross from Syria as Baghdad prepares its forces just one day after the U.S. military withdrawal.
Source: Fox News (lean-right, foxnews.com)
EXCLUSIVE: 128 GOP lawmakers call for federal probe into Planned Parenthood's taxpayer funding from Fox News
A coalition of 128 Republican lawmakers is urging the HHS Inspector General to review Planned Parenthood's federal funding for potential fraud or abuse.
Source: Fox News (lean-right, foxnews.com)
A must-win midterm state is the real loser of Trump's Canada trade war from Fox News
President Donald Trump's tariffs on Canada could raise vehicle prices at Michigan dealerships as auto parts cross the border multiple times.
Source: Fox News (lean-right, foxnews.com)
NY GOP Rep. Mike Lawler leads Dem Cait Conley in key battleground House district: new internal poll from New York Post
The incumbent GOP pol leads with 50% of the vote compared with Conley's 47% in the 17th Congressional District battle, according to the Lawler campaign's internal poll of 400 likely voters.
Source: New York Post (lean-right, nypost.com)
Here’s the real reason President Trump invited all those AI moguls to lunch this week from New York Post
Trump knows a lot is on the line as House Speaker Mike Johnson has been sounding the AI alarm bells as they relate to the GOP's chances in the 2026 midterms.
Source: New York Post (lean-right, nypost.com)
Super PACs prop up Hochul re-election, aiming for big win that will fend off ‘radical’ Dems from New York Post
Super PACs backing Democrat Kathy Hochul are opening up the spigot to aid her re-election bid against Republican rival Bruce Blakeman.
Source: New York Post (lean-right, nypost.com)
Roy Cooper steered $100M in NC disaster relief to low-income housing as hurricane victims went without homes from New York Post
After two devastating hurricanes, the Democratic then-governor's admin steered some $100 million in disaster funding toward affordable housing projects despite hundreds of storm victims stuck in temporary homes, according to an analysis.
Source: New York Post (lean-right, nypost.com)
Rep. Jayapal doubted Seattle Mayor Katie Wilson’s leadership over mass shooting response in texts with governor from New York Post
Progressive House Rep. Pramila Jayapal (D-Wash.) doubted Seattle Mayor Katie Wilson's leadership as officials scrutinized her response to the deadly mass shooting at a food festival.
Source: New York Post (lean-right, nypost.com)
The 1990 midterm elections offer survival tips for Republican candidates in 2026 from Washington Examiner
Republicans are headed for a rough reckoning next month. Persistent inflation, record fuel prices, and widespread disapproval of the war with Iran have soured the outlook for the GOP. According to the latest Wall Street Journal poll, President Donald Trump’s 37% approval rating is the lowest for any president since 1990. That comparison rang alarm […]
Source: Washington Examiner (lean-right, washingtonexaminer.com)
Liberals hope California ballot measure will be the foot in the door for wealth taxes from Washington Examiner
Californians will decide this November whether to impose a historic one-time wealth tax on billionaires. If they are successful, proponents hope to launch similar efforts across the country to tax the wealth of the ultra-rich. Passage of Proposition 40 would enact the first state-level wealth tax, which the ballot measure’s backers, including socialists, see as […]
Source: Washington Examiner (lean-right, washingtonexaminer.com)
El Sayed’s comparison between 9/11 attacks and U.S. response is ‘absolutely offensive’: Byron York from Washington Examiner
Washington Examiner chief political correspondent Byron York blasted Michigan Democratic Senate nominee Abdul el Sayed for his comparison of the Sept. 11, 2001, terrorist attacks and the United States’s response. “The idea of equalizing what was done on 9/11 by al Qaeda and what was done afterwards by the United States and other Western powers […]
Source: Washington Examiner (lean-right, washingtonexaminer.com)
Kristi Noem files for divorce from husband Bryon after ‘irreconcilable differences’ from Washington Examiner
Former Secretary of Homeland Security Kristi Noem filed for divorce from her husband Bryon Noem after 34 years of marriage. The divorce was filed due to “irreconcilable differences,” according to documents obtained by Dakota News Now. Kristi Noem said she was “devastated” after allegations surfaced that her husband was a secret crossdresser and chatted online […]
Source: Washington Examiner (lean-right, washingtonexaminer.com)
A captivating new biography of the flawed fixer Roy Cohn from Washington Examiner
When Roy Cohn died of AIDS in 1986, Donald Trump was among the few to attend his burial at Union Field Cemetery in Queens. He also attended Cohn’s public memorial later that year in Manhattan, along with Rupert Murdoch, George Steinbrenner, Barbara Walters, and William F. Buckley. Apart from his legions of well-connected friends and […]
Source: Washington Examiner (lean-right, washingtonexaminer.com)
Exclusive: New church campaign says America's family crisis won't be solved by politics, but by pastors from Fox News
Communio's JP De Gance says unmarried people in their 30s are the loneliest demographic and urges churches to actively encourage marriage now.
Source: Fox News (lean-right, foxnews.com)
FAA breaks controller hiring record after chief warned system was ‘chronically understaffed’ from Fox News
The Federal Aviation Administration hired 2,235 air traffic controller trainees this fiscal year, smashing its all-time record amid staffing shortages.
Source: Fox News (lean-right, foxnews.com)
Death of the Dinner Party from The Atlantic
The share of adults who say they regularly have friends over has declined dramatically since 1975.
Source: The Atlantic (unrated, theatlantic.com)
Inside Larry Ellison's Wide Web of Power, Money and Politics, As Oracle's Data Centers Claim Force Majeure from Vanity Fair
The $200 billion software titan has suctioned himself to many hubs of influence—from data centers to TikTok to Hollywood studios to the Oval Office itself. Tom Dotan dives deep into Ellison’s kingdom, starting on the island enclave of Lanai.
Source: Vanity Fair (unrated, vanityfair.com)
The Next Front from The Atlantic
The Baltics are getting ready for a Russian attack.
Source: The Atlantic (unrated, theatlantic.com)
Democratic Michigan Senate hopeful Abdul El-Sayed explains how science influences his politics from Scientific American
Scientific American spoke to the Michigan Senate candidate about how his background in medicine has shaped his outlook
Source: Scientific American (unrated, scientificamerican.com)
Republicans need Trump to turn out the base. Some candidates still don’t want him. - Politico from Slashdot.org
Republicans need Trump to turn out the base. Some candidates still don’t want him.Politico GOP breaks with Trump on multiple issuesThe Hill In Nine Key Senate Races, Five Republicans Won’t Commit to Campaigning With TrumpThe New York Times As Republicans brea…
Source: Slashdot.org (unrated, slashdot.org)
BRS Leaders Slam Revanth Reddy Over Sircilla Visit, Allege Diversionary Politics from The Times of India
BRS criticizes CM Revanth Reddy’s Sircilla visit, alleging diversionary politics and no new schemes; claims 54 weaver suicides since Congress took power.
Source: The Times of India (unrated, timesofindia.indiatimes.com)
Talarico launches $25 million statewide initiative targeting Black voters from Freerepublic.com
Democratic U.S. Senate candidate state Rep. James Talarico has launched a $25 million statewide initiative to reach Black voters six weeks before election day. The initiative is multifaceted and launched as Talarico continues to lead Republican Attorney Gener…
Source: Freerepublic.com (unrated, freerepublic.com)
Who is A'ja Wilson? WNBA star who confronted an Indiana Fever fan at a playoff game from Fox News
A'ja Wilson confronted a fan pointing at her during the Aces vs Fever game, earning a technical foul but standing firmly behind her reaction.
Source: Fox News (lean-right, foxnews.com)
Science News
A new physical model proposes that Venus’ mysterious haze is composed of cosmic dust and sulfuric acid, not unknown chemistry, from Ars Technica Science
Researchers built a physical model to explain the long-unidentified haze layer hanging above Venus’ clouds. The model matches observations by combining iron-rich cosmic dust with sulfuric acid droplets already known to exist on the planet.
. That combination yields the right light-scattering properties, solving a decade-old puzzle without invoking exotic particles.
. The work links an atmospheric mystery to an external source — micrometer-scale dust continually falling into Venus from interplanetary space.
.
The findings sharpen the picture of how Venus processes material from its environment, rather than only through internal chemistry
Source: arstechnica.com
SpaceX resolved a medical issue hours before launch, then flew its Crew-13 mission with Jessica Watkins as the first Black woman to command a spaceflight, from Ars Technica Science.
The Crew-13 mission lifted off after SpaceX performed a "surgical intervention" on a technical problem detected in the final countdown. Engineers isolated and corrected the fault without rolling the rocket back, keeping the launch window intact. The flight proceeded to orbit with the four-person crew aboard a Crew Dragon capsule atop a Falcon 9.
Jessica Watkins served as mission commander, making her the first Black woman to lead a crew into space. The milestone comes after her previous role as a mission specialist on Crew-4, where she also became the first Black woman to complete a long-duration stay on the International Space Station. NASA and SpaceX officials confirmed the crew is healthy and the capsule is on track for docking.
The intervention itself was not a safety compromise, according to SpaceX's post-launch briefing. The issue was isolated to a non-critical ground system component, allowing the team to bypass it with a redundant path. Watkins and her crewmates are now beginning a roughly 24-hour transit to the station, where they will join the current Expedition crew for a six-month science rotation.
Source: arstechnica.com
AI finally cracked the hidden-information game Stratego by adding a second neural network to guess opponent piece identities, from Ars Technica Science.
Stratego, a board game where all pieces are hidden, had long resisted AI because players must infer identities from movement patterns. Previous systems relied on a single network to reason over incomplete information, but they kept making brittle, easily-exploited assumptions.
A team broke through by splitting the problem: one network handles move planning, while a second one explicitly estimates the probability of what each hidden enemy piece actually is. That second network feeds its guesses back into the first, sharpening decision-making over time at each turn.
The new algorithm was tested against Stratego AI baselines and held its own against top human players. It didn’t need perfect knowledge—the key was learning to live with uncertainty by constantly revising its model of the board. This mirrors how humans play, and it suggests the approach could port to other imperfect-information games like poker or even real-world negotiation scenarios, where bluff and deduction matter more than brute computation.
Source: arstechnica.com
Following the MIND diet more closely was linked to slower brain shrinkage and less gray matter loss, with effects equivalent to about 2.5 years of delayed brain aging, from Science Daily.
The study tracked participants over time and compared their diets against the MIND score. People with higher adherence showed notably less brain tissue loss, and the strongest differences translated to roughly two and a half years of preserved brain health. This association was independent of other lifestyle factors.
.
Berries and poultry stood out as the foods most tied to healthier brain changesbi. Conversely, sweets and fried fast foods correlated with greater tissue loss. The pattern suggests specific food choices, not just overall diet quality, may drive the effect. No foods were framed as a silver bullet—the signal emerged from the whole dietary pattern, with certain items pulling the average either way
Source: sciencedaily.com
Researchers found the Alzheimer’s risk gene APOE4 may damage brain blood vessels and impair protein clearance, with some effects reversible in experiments, from Science Daily
The study identifies APOE4 as an active disruptor of blood-brain barrier integrity, not just a passive risk factor. This vascular damage appears tied to the same cellular machinery that normally flushes out toxic protein buildup — theggested shared mechanism across Alzheimer’s, Parkinson’s, and other neurodegenerative conditions.
In lab experiments, the team successfully reversed several of these effects, restoring function in damaged cellular systems. That reversibility points to specific molecular pathways as viable therapeutic targets, though the work remains preclinical and far from clinical application.
The findings reframe APOE4’s role from genetic predisposition to active pathological agent, potentially explaining why carriers face earlier or more aggressive disease onset. If vascular injury proves reversible in living systems, it could open a treatment angle that addresses disease progression rather than symptom management alone.
Source: sciencedaily.com
A small preliminary study found that people with depression who sharply reduced ultra-processed foods saw notable symptom improvements, from Science Daily.
The trial enrolled people with depression and had them make major cuts to ultra-processed foods. Over the study period, their depressive symptoms improved measurably — a signal, not proof, given the small sample and lack of control group.
Researchers behind the work think the effect may run through the gut. They suspect that ultra-processed foods damage the intestinal lining and feed chronic inflammation, which in turn can affect the brain. Cutting them out, they argue, may quiet that inflammatory pathway.
The findings are early, and the study wasn’t designed to isolate whether the improvement came from removing junk food or from eating more whole foods instead. Still, it adds to a growing pile of evidence linking diet quality to mood — and suggests that for some people, food choices might be a lever worth pulling alongside standard depression care.
Source: sciencedaily.com
.jpg){kind=link}