OpenAI starts testing ads inside ChatGPT answers

OpenAI begins slipping ads into ChatGPT for U.S. users, pitching it as a way to fund wider access. Readers worry that a trusted assistant is turning into a targeted marketing funnel, with paid plugs quietly shaping what people see as neutral advice.

Teen overdose linked to ChatGPT drug advice

A California teenager reportedly dies after following ChatGPT guidance about kratom and other drugs, laying bare how human-sounding bots can give confident, deadly nonsense. Many see this as proof that shiny LLMs are being shipped into the world without real guardrails.

AI-generated hit song banned from Swedish charts

A catchy track is kicked off Sweden’s official charts solely because it is an AI creation, not a human act. Fans and creators argue whether chart rules are protecting culture or just gatekeeping, while AI music suddenly feels less like a fun toy and more like a legal fight.

Law scholar warns AI wrecks key institutions

A soon-to-be-published law article, "AI Destroys Institutions", argues that powerful AI systems will quietly hollow out courts, media, and markets. The community reads it less as sci‑fi and more as a grim manual for what happens when profit-chasing models rewrite the rules.

Why we forgive AI mistakes more than humans

An essay claims people cut AI endless slack while jumping on human errors, because we treat bots like tools and ourselves like the problem. Many nod along, seeing it in daily use of assistants that hallucinate, while humans get blamed for “bad prompts” instead of bad design.

U.S. cyber strike in Venezuela shows silent firepower

A deep dive into a 2019 U.S. cyberattack on Venezuelan air defense radar shows how software, not missiles, can blind an opponent. The story confirms that modern cyberweapons are precise, deniable, and political, and readers wonder how often this happens without headlines.

German leader brands nuclear shutdown a huge mistake

Conservative leader Friedrich Merz blasts Germany’s exit from nuclear energy as a “huge mistake,” arguing it drove up prices and emissions. With renewables and gas still in a messy balance, tech-minded readers see a textbook case of how long-term energy planning can misfire.

Study says renters locked out of energy savings

New research from Binghamton University finds most renters can’t access weatherization and efficiency upgrades, even as heating costs soar. The work highlights a landlord–tenant deadlock that leaves poorer households freezing while policy talk about green tech stays abstract.

Canada slashes tariffs on Chinese electric cars

Canada cuts tariffs on Chinese EVs from 100% to 6%, opening the door for cheap models like the BYD Seagull. The move thrills budget‑minded drivers but feels like a grenade under North American auto and battery plans, with politics and climate goals colliding hard.

Mandiant drops rainbow tables to kill old auth

Security firm Mandiant releases Net‑NTLMv1 rainbow tables, making it dramatically easier to crack this ancient protocol and forcing holdouts to move on. Admins call it both a public service and a wake‑up slap, proof that lazy legacy setups are now a glaring liability.

Astro web framework moves into Cloudflare empire

The team behind Astro joins Cloudflare, folding a beloved static-site framework into a massive edge network. Some devs cheer the scale and funding, others worry that every promising open source tool eventually gets a corporate logo and a long roadmap of lock‑in.

Just the Browser strips Chrome of noisy bloat

An open project called Just the Browser helps users gut modern browsers of AI fluff, telemetry, sponsored junk, and nagging integrations. The response feels almost cathartic, as people admit they mainly want a fast window to the web, not a clingy ad platform on their desktop.

Cursor accused of overhyping coding agent experiments

A detailed post claims Cursor marketed its latest “autonomous coding” browser experiment as a big success without sharing solid evidence. Devs are increasingly tired of puffed‑up AI demos and demand messy, real benchmarks rather than cherry‑picked screenshots and vibes.

Founders debate if software startups still make sense

In an Ask HN thread, would‑be founders question whether starting a software startup is still worth it in an AI‑flooded, Big Tech‑dominated world. Replies swing between doom and opportunity, but everyone agrees that easy wins are gone and niches now matter more than buzz.

DuckDB wins hearts as simple local data powerhouse

A long‑time user explains why DuckDB has become their go‑to tool for crunching data on a single machine instead of firing up heavy cloud stacks. The story taps into a strong mood: devs are tired of overbuilt pipelines and love tools that are fast, boring, and local.

Claude workbot tricked into leaking private files

A security researcher shows Claude Cowork can be steered by crafty prompts into reading and exfiltrating files from its own workspace. People who just wired these agents into real code and documents suddenly feel exposed, and trust in neat "secure by design" claims takes a real hit.

Anthropic quietly blocks OpenCode from Claude API

Without real warning, Anthropic cuts off the popular OpenCode tool from Claude Code. Devs see it as the careful, safety‑first company playing platform cop, and the reaction is swift: angry posts, promises to try rivals, and fresh worries about building businesses on someone else’s model.

Community says Anthropic makes a huge mistake

After the OpenCode block, a wave of users call Anthropic’s move self‑defeating and petty. The tone is weary: yet another AI giant treating partners as disposable. Many say this is the push they needed to test open models instead of staying locked to one polished corporate brain.

Autonomous coding agents now run for weeks alone

The Cursor team shares experiments where AI coding agents hack on projects for weeks with minimal human input. It sounds futuristic, but the write‑up admits constant failures, messy code, and lots of babysitting. Readers feel both impressed and uneasy about what this means for everyday programmers.

Junior devs fear there is no entry door

A blunt essay on junior developers in the AI era argues companies like the idea of replacing beginners with tools. Paired with a raw rant from an unemployed MIT grad, it captures a heavy mood: older engineers worry for newcomers, and newcomers wonder if the ladder has already gone.

Fortnite maker fined for nudging kids to pay

A Dutch court confirms a long‑running Epic Games fine of over one million euros for how Fortnite pushed children toward in‑app purchases. Gamers cheer the slap, parents nod along, and designers quietly recheck every bright button and countdown timer that keeps the money flowing.

UK backs off plan for mandatory work digital ID

The British government drops a plan to make a new digital ID the default proof of right to work, returning to passports and existing checks. Privacy‑minded readers breathe out, seeing it as a rare win against creeping ID systems that turn every job hunt into a database check.

EFF explains how to survive new age gates

With more sites rolling out hard age gating rules, the EFF walks users through what these laws do, the risks of face scans and private IDs, and how to push back. The tone is practical but worried, as online life starts to feel more like a bouncer line than an open web.

Push grows to ban social media under sixteen

A long piece argues every country should set 16 as the minimum age for full social media accounts, pointing to mental health and addiction data. Many parents nod, teens roll their eyes, and tech workers brace for yet another wave of rushed rules written by people who barely post.

DHS deportation clips hit by music copyright claims

The US DHS posts slick deportation reels on social media, only to see them slapped with copyright strikes for unlicensed music. Critics call it a perfect symbol of the moment: a government agency eager for viral punishment clips but apparently too sloppy to clear the soundtrack.

New Linux cloud malware looks painfully professional

Researchers detail a Linux malware family hitting cloud servers with stealthy tricks well above script‑kiddie level. It slips through common defenses and abuses provider features, leaving admins unsettled and yet again questioning how many ghosts are already living in their containers.

Starlink roam data doubled with slow unlimited after

SpaceX quietly bumps Starlink Roam from 50 to 100 GB of high‑speed data, then unlimited but slower traffic after that, at the same price. Van lifers and rural users seem pleasantly surprised for once, though everyone is waiting to see what "slow" really feels like on movie night.

Hackers, Iran and Starlink GPS spoofing tests

A deep dive into Starlink terminals in Iran shows how they detect GPS spoofing and jamming attempts. It reads like spy fiction with code samples, and leaves readers impressed that satellite dishes now play cat and mouse with nation states before they even load a single web page.

Verizon outage leaves US east coast scrambling

A major Verizon outage hits parts of the US east coast, pushing people onto Wi‑Fi calls and iPhone emergency satellite features. The story feels familiar and tired: one big provider stumbles, support scripts lag, and customers are reminded how fragile their always‑online life really is.

New URL shortener leans into maximum sketchy vibes

A joke project offers a URL shortener that makes links look as suspicious as possible on purpose. It is silly and sharp at the same time, poking fun at how used to shady redirects we have become and how little comfort a clean looking link truly gives anyone these days.

One Nation Vanishes From The Internet Overnight

Iran’s regime hits the kill switch and 90 million people suddenly lose the internet for days. The report tracks collapsing BGP routes, silent mobile networks and people scrambling for VPNs and offline tools. It feels more like a dress rehearsal for digital control than a brief outage.

UK Wants Phones To Spy Before Crimes Happen

New plans under the UK Online Safety Act push for mandatory, automated scanning of private messages and tools like AirDrop. Supporters say it protects children, but critics see an open door for mass surveillance, backdoors and mission creep. It reads less like safety and more like precrime by design.

Britain Builds A Real Life Precrime Machine

This piece connects tougher protest laws with algorithms, facial recognition and predictive policing pilots across the UK. The mood is grim: once you normalize scanning crowds and guessing who might offend, the tech rarely stays limited. The line between public safety and dissent control looks thinner every month.

Signal Calls Agentic AI A Security Disaster

The privacy-first Signal team tears into "agentic" AI baked into operating systems. They warn about OS-level recorders like Recall, giant life-logs ripe for malware and people being opted in by default. It is a blunt reminder that convenience assistants can quietly become the ultimate surveillance layer.

Security Pros Say AI Makes Defenses Weaker

A seasoned security voice argues that AI will wreck defenses not by movie-style superintelligence, but by turbocharging phishing, password guessing and social engineering. Tools like PassGAN already chew through weak logins. The piece drips with frustration at managers who buy AI hype while underfunding basic hygiene.

Coal Roars Back As AI Power Bills Soar

Fresh numbers show US emissions rising 2.4 percent in 2025 as coal plants spin up again and AI data centers gulp electricity. Solar and wind keep growing but cannot yet outrun demand. Commenters sound tired of hearing about a green future while real-world carbon charts bend the wrong way.

Apple Bundles Pro Apps Into Creator Playground

Apple unveils Creator Studio, a bundle pulling Final Cut Pro, Logic Pro and friends into one glossy package. Fans see a creative playground, skeptics see another move to lock artists into Apple hardware and subscriptions. Either way, the company is clearly courting the booming creator economy.

US Drone Ban Hits DJI And Hobby Flyers

The US finally follows through on its DJI crackdown, blocking key approvals and effectively grounding new Chinese-made drones. Officials cite security fears, but firefighters, filmmakers and hobby pilots fear higher prices and worse gear. It is another messy collision of geopolitics with everyday technology.

Anthropic Pours Millions Into Python’s Nerve Center

Anthropic pledges 1.5 million dollars to the Python Software Foundation, money aimed squarely at core development and PyPI security. For a community used to scraping by, the gift feels huge. It also quietly admits that billion-dollar AI labs still balance on top of fragile open source plumbing.

Sam Altman’s Star Dims As Rivals Catch Up

Gary Marcus lays out how OpenAI and Sam Altman went from untouchable to merely early as GPT-5 drags, Apple cozies up to other partners and competitors ship fast. The piece voices a growing sense that the AI race is becoming a grind, not a coronation for a single hero.

Tired Gadgets Should Reveal Their Secret Code

A short manifesto argues that when hardware hits EOL, companies should be forced to open-source the software so users can keep devices alive. The idea taps into anger at smart speakers and headphones bricked by corporate neglect and leans on growing "right to repair" momentum.

AI Scrapers Are Wrecking Beloved Music Databases

The MetaBrainz crew details how rogue AI scrapers hammer sites like MusicBrainz, ignoring robots.txt and rate limits. The tone is exhausted and angry: volunteers build free cultural archives, and faceless labs quietly strip-mine them for training data without even asking. It feels like theft dressed as progress.

Indie Game Studio Proudly Refuses Any AI Help

The team behind Yarn Spinner flatly says "we do not use AI" for writing or art and explains why. They worry about training on stolen work and losing human voice. Readers resonate with the honesty, seeing it as proof that small studios can still choose craft over quick content.

Developer Warns Chat Interfaces Are UX Dead End

A veteran developer argues that natural language interfaces are overused, slow and expensive. For many tasks, buttons and forms beat chatting with a bot. The take lands with devs tired of bolting LLMs onto everything just for hype, and it hints at a coming backlash in product design.

Teardown Shows Smart Tech Hiding In Trashy Vape

A curious hacker rips apart a discarded disposable vape and finds a USB-C port, a neat LiPo battery, a capable microcontroller and tidy circuitry. The teardown makes cheap throwaway gadgets look disturbingly advanced, and the waste of perfectly good electronics leaves readers both impressed and annoyed.

New license tells hungry AI to back off

This new MIT Non-AI License bolts a big keep out sign on open-source code for commercial AI training. Many developers cheer the pushback, others worry it breaks the spirit of sharing and will be nearly impossible to enforce in the real world.

Burned-out maintainer slams door on open-source

After years giving away open-source work, this developer slams the door and vows to ship only closed code. The trigger is big companies and AI tools profiting off unpaid labor, a sore spot many coders say they feel deep in their bones.

AI turns shaky business plans into smoke tests

This essay argues AI is a stress test for every business model. Anything you can fully describe, chatbots can copy for pennies, but messy, hands-on services still hold value. It reads like a quiet warning label for startups selling thin wrappers.

Live AI models leak entire books on demand

Researchers show how production AI models can spill entire books they were trained on, just by poking them the right way. It’s a nightmare for publishers and a smoking gun in the fight over whether training data is really forgotten inside the model.

ChatGPT Health makes patients the real product

ChatGPT Health launches as a shiny health assistant, but the fine print turns patients into the product. It plugs into apps like Apple Health, raising sharp questions about consent, data sharing, and what happens when your symptoms feed ad engines.

Iran runs chillingly precise internet shutdown on protests

Iran’s rulers flip a surgical internet kill-switch, blocking protesters while keeping government systems alive. The blackout is scarily precise, and many see it as a dress rehearsal for how future regimes might silence dissent without crashing business.

UK pushes Ofcom toward risky chat backdoors

Britain orders Ofcom to explore ways to scan encrypted chats for abuse, which likely means backdoors. Privacy defenders call it a dangerous fantasy that weakens security for everyone, while politicians keep pretending there is a safe magic workaround.

UK lets itself dodge its own cyber rules

A new cyber law in the UK lets government bodies sidestep some of the same rules they expect from everyone else. After a string of hacks, it feels backwards, and critics say it sends one clear message: accountability is for you, not for them.

CDC staff blindsided by sudden vaccine rule overhaul

A surprise move from the Trump administration rips up the child vaccine schedule with little input from career scientists. CDC staff say they were blindsided, and the whole thing looks like another round of politics steamrolling careful public health work.

USDA freezes Minnesota funds after fraud scandal

The USDA suddenly freezes federal funds to Minnesota and Minneapolis after a huge food aid fraud scandal. Officials say enough is enough, but locals fear kids and low-income families will pay the price while the political blame game drags on.

Datadog ban helps rival sell its own product

A small startup using Datadog for monitoring gets its account flagged and then frozen, right as it builds a cheaper rival on OpenTelemetry. The team turns the mess into a fiery blog post that doubles as an ad for ditching pricey, locked-down dashboards.

One VM, a hundred Linux dev shells, no hype

This Show HN project squeezes a hundred Linux dev environments onto one virtual machine using old-school containers instead of trendy Kubernetes. Programmers love the simplicity and low cost, and it quietly shames how bloated many modern cloud setups feel.

Fly.io says short-lived sandboxes are yesterday’s news

Fly.io declares old-school short-lived sandboxes dead and pushes long-running app sandboxes instead. The idea is to keep tiny machines alive to reuse warm state and cut cold starts. Fans see clever engineering, skeptics see yet another flavor of cloud lock-in.

Ghostty terminal hunts down monstrous memory leak

Popular terminal app Ghostty turns into a memory hog, with one user seeing it eat 37 GB. The maintainer hunts down a sneaky leak and writes a detailed post-mortem. Devs admire the detective work and quietly panic about hidden bugs in their own tools.

Arch Linux package guts get a Rusty upgrade

A year-long push funded by the Sovereign Tech Fund modernizes Arch Linux’s package manager, shifting key pieces into Rust. It’s unglamorous plumbing work, but users cheer, because safer, well-funded infrastructure beats flashy features that break updates.

Iran’s rulers drag internet down to one percent

A nationwide internet shutdown in Iran has dragged on for 24 hours, dropping connectivity to about one percent of normal. It feels less like a security move and more like smashing the public square, with every outage turning tech into a blunt political weapon.

DHS leans on immigration rules to grab DNA

DHS is using immigration enforcement as the doorway to scoop up DNA from people, including Americans, building a biometric stockpile that looks far bigger than any border problem. It comes across as mission creep turned permanent, with privacy left in the dust.

Flock Safety exposes key to license plate empire

Police-tech vendor Flock Safety hardcoded an ArcGIS API key more than 50 times in public pages, apparently exposing access to layers tied to license-plate scans and crime data. It feels like a bad joke: mass surveillance sold as safety, guarded with copy‑paste security.

Cloudflare blasts Italy over rushed blocking orders

Italy hit Cloudflare with a $17M fine for refusing a scheme to block Olympic piracy traffic within 30 minutes. The CEO paints it as a dangerous shortcut to censorship, and the whole fight sounds like a preview of how messy future net policing will get.

ICE detains thousands who have no convictions

New numbers show about 73% of people detained by ICE have no criminal convictions at all. Put next to DNA dragnets and expanded tracking, immigration starts to look less like border control and more like a convenient cover story for building a giant monitoring machine.

Grok kills images after sexual content outrage

Elon Musk’s Grok turns off its image generator for most users after regulators and critics slam it for sexualised pictures, including of public figures. It feels like AI is sprinting ahead, then smacking into a wall of rules and public anger almost overnight.

AI helps crack a classic Erdős math puzzle

An AI system, with light human feedback, helped solve Erdős problem #728, a long‑standing challenge from the famous mathematician. It lands like a plot twist: the same tech that writes boilerplate code now pushes into serious mathematics, blurring where human insight ends.

Open-source tool checks EU AI Act compliance offline

EuConform arrives as an open‑source, offline EU AI Act compliance helper that flags risk levels and bias. It reads like a survival kit for smaller teams staring at new rules, trying to stay legal without handing all their data to yet another online scanner.

Developer calls new AI integration standard a fad

A sharp take brands Model Context Protocol (MCP) as overhyped glue code for AI tools, not a revolution. The criticism taps into a growing mood: every big player slaps out a standard, but devs are tired of rearranging their stacks for trends that may not last.

Engineer warns against blind faith in coding AI

A senior dev praises using AI assistants but slams the cult around them, warning that teams risk losing design sense and basic discipline. The piece mirrors a quiet worry: tools like Claude Code and Cursor help, but they also tempt people to stop thinking deeply.

Android source now drops only twice a year

Google’s new AOSP policy waits until Q2 and Q4 to publish source, after releases. For many builders, this feels like Android edging away from true open source, turning community devs into late guests at a party they helped decorate in the first place.

EU hunts for open-source power to escape Big Tech

The European Commission launches a call for evidence on open source as it sketches a European Open Digital Ecosystem. It reads like Brussels trying to kick a dependency habit on non‑EU software and cloud, with coders hoping it means real funding, not just speeches.

Kagi’s Orion browser finally lands on Linux alpha

Kagi ships an alpha of its Orion browser for Linux, giving privacy‑minded users another alternative to the usual giants. It is buggy and early, but the excitement shows how hungry people are for new engines that are not controlled by ad empires.

Writer argues your shell does not need Oh My Zsh

A popular post claims Oh My Zsh is bloated, slow and unnecessary, urging users to hand‑pick a few plugins instead. The tone fits a wider backlash against heavyweight tooling, where people are tired of waiting seconds for a terminal that used to snap open instantly.

Study shows cloud hardware gains slowing, costs rising

The Cloudspecs paper digs through a decade of cloud hardware data and finds network speeds soaring but core performance gains flattening. It feeds a nagging feeling that the easy days of ‘infinite scale’ are fading, while vendors quietly nudge customers toward pricier tiers.

Google attacks bot that scrapes its results

Google sues SerpApi, accusing it of dodging locks to slurp search result pages, including rich knowledge panels. Many developers see this as a warning shot at scraping in general and worry that big platforms want to fence off more of the public web.

Gmail turns into full blown AI secretary

Google plugs its Gemini model straight into Gmail for 3 billion users, promising smart drafts, summaries and follow up suggestions. Fans dream of an inbox that finally calms down, while skeptics shiver at one company’s AI reading and reshaping almost every mail on Earth.

Nvidia unveils monster chips for next AI wave

Nvidia shows off its Rubin platform, bundling new GPUs, CPUs and a giant AI supercomputer aimed at the next generation of massive models. The move reminds everyone how dependent modern AI labs are on one chip giant, and how hard it will be for rivals to catch up.

Coders say AI helpers now feel half asleep

Developers complain that AI coding assistants which once felt sharp now spew more nonsense, hunt for upsells, and miss simple patterns. The grumbling hints that model quality may have plateaued while vendors chase growth, leaving many wondering if the magic is slowly wearing off.

IBM coding bot caught running stranger’s malware

Security researchers show IBM’s Bob coding agent can be tricked into downloading and running malware through sneaky prompts, with no human double check. It is a jarring example of how fast helper bots can turn into attack tools when guardrails assume every command is friendly.

Iran cuts internet as street protests spread

Network watchers report Iran sharply dropping off the global internet map as protests grow, with traffic plunging at major providers. The blackout shows again how quickly a government can seal digital borders, leaving people scrambling for any remaining link to the outside world.

US border cops quietly track phones by neighborhood

Leaked slides reveal ICE buying tools called Tangles and Webloc that can watch phone activity by area and follow devices back to homes, often without warrants. It confirms what many feared: location data sold by private firms has become a ready-made dragnet for government.

Odd internet glitch in Venezuela sparks outage fears

A Cloudflare analysis digs into a strange routing move by Venezuela’s main provider that briefly shoved web traffic onto an unusual path. The incident fuels concern that fragile internet plumbing and missteps by big carriers can silently knock whole regions offline at any moment.

Texas kicks bar association out of law schools

Texas leaders vote to end ABA oversight of state law schools and set up their own rules. Critics see political payback and worry about lower standards, while supporters cheer a chance to mold a homegrown pipeline of lawyers more in line with local views.

Samsung busted over TVs that watch the watchers

A Texas court briefly blocks Samsung from using tracking tech in its smart TVs, then quickly pulls back the order. The whiplash leaves viewers uneasy, since it confirms sets can quietly log what people watch while the legal system struggles to draw a clear line.

Bose frees old speakers instead of killing them

Facing aging SoundTouch gear, Bose chooses not to brick the speakers but to open up software so enthusiasts can keep them alive. It is a rare corporate move that treats customers like owners, not renters, and it instantly wins goodwill from gadget lovers tired of forced upgrades.

Clip on screen turns MacBook into drawing tablet

A project called Intricuit straps a touch layer onto a MacBook screen and ships with a pressure sensitive stylus, giving laptops tablet style sketch powers. It scratches that itch many feel for pen input without buying yet another glowing slab just to doodle or mark slides.

Classic Casio watch upgraded into secret pay gadget

A hacker squeezes NFC payment tech into the humble Casio F-91W, turning a cheap digital watch into a tap to pay wrist wallet. Fans love seeing a mass market classic gain sci fi tricks, and it highlights how much hidden room still sits inside everyday plastic shells.

One expired certificate bricks Logitech Mac apps

A simple expired certificate leaves Logitech’s Mac apps unable to run or even update themselves, wiping user settings in the process. The mess reminds everyone how fragile modern software chains are and how a tiny date field can strand thousands of pricey mice and keyboards.

New Rust toolkit promises calmer tiny gadgets

The Embassy project pushes a modern Rust framework for small devices, aiming to make low power gadgets safer and easier to program. Embedded fans are hopeful it will end years of flaky firmware and random freezes that make smart toys and sensors feel dumber than their ads.

Congress aims straight at H‑1B tech visas

A new bill to kill the H-1B visa program lands like a bomb in Silicon Valley, threatening thousands of skilled immigrant workers and the companies that depend on them. Supporters call it protection for locals, critics see political grandstanding that will only push talent overseas.

US targets Wall Street home hoarders at last

The US plans to ban big Wall Street firms from buying single-family homes, blaming them for tight supply and rising rents. Homebuilder stocks wobble while renters cheer, but many doubt regulators will really unwind years of financialization or stop investors from finding new loopholes.

ICE splashes $28B on new surveillance toys

With a massive $28.7B budget, ICE goes shopping for databases, phone trackers and other surveillance tech. Civil liberties watchers are alarmed, seeing immigration enforcement turning into a general monitoring platform that can quietly track almost anyone, not just people at the border.

Greenland’s melting ice hides a mineral goldmine

New reporting on Greenland details huge untapped mineral deposits under retreating ice, from rare earths to metals vital for green tech. Locals fear a fresh resource rush that trades one climate problem for another as mining giants eye the Arctic like a new Wild West.

The $14 burrito shows why inflation still stings

A deep dive into San Francisco’s $14 burrito explains why official inflation numbers feel fake to locals. Tech workers and baristas alike see daily prices that never fall, even as statistics say things are calm, feeding suspicion that the system is tuned to soothe markets, not people.

ChatGPT Health reaches into your medical records

OpenAI launches ChatGPT Health, promising smarter answers by mixing AI with real health records. The idea is seductive, but people are worried about leaks, bias and who gets blamed when a slick chatbot gives the wrong call on a diagnosis or drug and doctors are tempted to trust it.

Notion AI can leak data before you approve

A researcher shows Notion AI is open to indirect prompt injection, saving poisoned edits to documents before users hit OK. It turns a friendly writing buddy into a sneaky data exfiltration tool, and the fact it stays unpatched leaves teams wondering what else their AI helpers log.

Tailscale quietly drops default state encryption

A Linux update from Tailscale removes automatic state file encryption and loosens its hardware checks so the client can start more easily. Fans of the service feel uneasy, since a product sold on security just made a tradeoff that leaves sensitive config data sitting easier to read.

Linux kernel bugs lurk for decades unnoticed

New analysis of Linux kernel history finds bugs hiding in code for an average of two years, with some sleeping for twenty. It is a sobering reminder that the machinery behind phones, servers and routers is full of latent vulnerabilities that nobody spots until luck or disaster strikes.

Popular JS crypto library ships with hidden flaws

Security firm Trail of Bits uses Google’s Wycheproof tests to uncover vulnerabilities in the widely used elliptic JavaScript crypto library. With millions of weekly downloads, the news rattles developers who assumed the math was safe and now must wonder what secrets rode on weak code.

GNOME moves to kill classic Linux middle-click paste

A GNOME developer pushes to remove middle-click paste from modern desktops and even Firefox, enraging long-time Linux users. To fans, it feels like yet another case of designers sanding off powerful, weird traditions in the name of safety while ignoring what actually made the platform fun.

Firefox add-on dodges X login wall with xcancel

A small Firefox extension silently redirects x.com and old twitter.com links to xcancel.com, letting people read threads without logging in. It is a petty but satisfying act of resistance against a hostile platform, and many clearly enjoy taking back a tiny bit of control in their browser.

Tailwind creator admits brutal 75 percent layoffs

The team behind Tailwind reveals they cut 75% of their engineering staff, blaming a brutal market. Devs are stunned that one of the web’s hottest CSS tools is shrinking, reading it as a warning that even beloved frameworks are not safe when VC dreams meet cold subscription numbers.

Hackers build open hardware clone of Wacom tablets

Project Patchouli offers a fully open electromagnetic pen tablet design, from coil arrays to firmware, aiming to be a kind of DIY Wacom. Hardware tinkerers love the freedom, and many see it as a rare case where creativity beats vendor lock-in instead of living under another closed driver.

Everyone piles on OneDrive’s dark patterns and bugs

A blistering rant about Microsoft OneDrive calls it a file-sync service that nags users, hijacks defaults and sometimes even loses or deletes data. Commenters loudly agree, treating it as proof that big vendors will happily trade reliability and consent for a few more files in their cloud.