micahflee.com
Flock cameras are riddled with security vulnerabilities and hardcoded creds
x.com
We're sharing our new framework for tracking, investigating, and disclosing instances of model misalignment at OpenAI. The framework sets criteria and timeline…
commoncrawl.org
Common Crawl Data Stored on a Hugging Face Bucket
x.com
Our first ever stealth model is up on @CloudflareDev AI Gateway — Union Alpha! The sweet spot for the best price/performance ratio, this model is making waves…
arxiv.org
Breaking the 1.58-bit Barrier for Ternary LLMs
filipovski.net
Backups Aren't Simple
mimo.xiaomi.com
Xiaomi Mimo 2.6 live post-training dashboard
cosmicave.org
Why Does the Universe Expand?
arstechnica.com
macOS 27 Golden Gate – Review
arxiv.org
How good are frontier models at physics?
wsj.com
Fed Raises Rates for First Time in Three Years
krebsonsecurity.com
Data Broker Radaris Loses Domains in Privacy Fight
barndoor.ai
Barndoor acquires Diaphora, creators of open-source workflow runtime Frags
x.com
Grok Build now gets better the more you use it. It remembers conventions, decisions, and project facts across sessions. /memory to browse /dream to organize re…
businessinsider.com
A coffee shop owner used AI to make a menu poster. Then came the angry DMs
arxiv.org
Accurate Models of AMD Matrix Cores
reuters.com
Fed hikes rates as inflation worries push up bond yields
github.com
Show HN: Friday – Self-hosted persistent memory for AI coding agents (MCP)
rohanbansal.com
Training a 4B model to produce 81% faster query plans than Postgres
github.com
Reverse-engineered Jev-like model
euronews.com
Berlin-based Langdock moves its parent company from the US to Germany
news.ycombinator.com
Tell HN: An inside view of Montana's new biotech law
Montana passed a law called SB535. It builds on right-to-try (pre-approval access with informed consent, Phase 1 safety data etc.) but goes much further, fixing problems with those laws. Alex Tabarrok called it "the most important regulatory innovation in drug approval in my lifetime" ( ). Was posted here but died in /new ( ). The mods suggested this post. I contributed ideas to the law and am now implementing it through my company. I invested in biotech for years and watched companies struggle. I built the biotech ecosystem in Prospera as an alternative, concluded it was too early, and now think Montana is the best place to prove this. Why this exists: When FDA approves a bad drug, heads roll. When it delays a good one, the deaths are statistical and nobody gets blamed ("invisible graveyard"). So the incentive is overcaution, which is why the cost per approved drug has roughly doubled every 9 years for decades ("Eroom's Law"). Founders are in the "Valley of Death" around Phase 1: grants are no longer available, and commercial money wants assurance the drug will pass the next trial. Only ~10% of post-Phase 1 drugs get approved, but 68% of failed trials don’t stop because they found lack of safety or efficacy but commercial reasons (Williams et al., PLOS ONE 2015). Federal Right to Try and Expanded Access haven't fixed this. Federal reform is super-hard. What Montana allows: A physician can give an experimental treatment outside a trial if: it completed FDA Phase 1 under an active IND; a state-registered private review board (ETRB) approved the protocol; it's delivered at a state-licensed clinic; consent exceeds the federal standard & adverse events need to be reported. The key is: sponsors and clinics can charge. Why this time it's different: The risk-reward ratio is what's broken. Right to Try and Expanded Access don't let sponsors charge, so treating a patient is risk plus expense. Montana is the first state law where sponsors of IND-stage drugs can price in that risk. Trial recruitment today is a price-control system: per-patient cost is around $50-100k, typically has a ceiling upward on what it can pay patients ("undue inducement") and a floor downward (no profit, only at cost in RTT / EA). Montana removes both (I know this will lead to lots of debate, let’s have it.) So this is not a free for all, the additional liberties come with tough oversight. An ETRB is Montana's version of an IRB: safety review, consent, mandatory outcome reporting, and you can't withhold safety information from patients. That’s the truth-funding mechanism. What companies can do now: If you have a Phase 1 asset stuck in the Valley of Death: treat patients, negotiate payment, get real-world data, use it to sharpen your Phase 2/3 design. Disclosure: my company formed the first ETRB. The model is review fees, like an IRB; no equity in applicants, no payment by outcome; COI policy and board bios public; decision letters published with applicant consent; annual outcome report required. Objections: - Someone gets hurt? Same as trials and ordinary care: US legal system, legal recourse. - FDA shuts it down? They haven't said they won't, but historically FDA goes after grey-market clinics, not state laws; we're asking for safe harbor, but some companies aren't waiting. - Snake oil? Bad actors want to fly under the radar, and Montana makes that hard. What's needed: Biotechs with Phase 1+ assets willing to move before full FDA assurance, to build the evidence that gets the agency on board - the point is not to skip FDA, but to reduce the cost of data. And ex-FDA reviewers, IND operators, IRB members telling us where this breaks. Happy to answer anything.
opensource.googleblog.com
Vectorized and performance-portable Quicksort (2022)
x.com
You can just build (physical) things.
chuckgreenman.com
GitHub is having trouble counting things
temertymedicine.utoronto.ca
ER visits for gambling disorders doubled after expanded online gambling market
x.com
Claude Design, Claude Slides and Claude Docs also work inside Claude Code now. Ask for a design review deck or a UI mockup and point it at the actual files and…
x.com
What can you do with GPT-6 Astra?
linum.ai
Training Text-to-Image Models 3.6× Faster
claude.com
Claude Cowork and chat are now one Claude
x.com
Claude Cowork and chat are merging into one Claude. Ask a quick question or hand over a report, and Claude takes it from there, even after you close your lapto…
testing.googleblog.com
This Code Is CRAP (2011)
will-keleher.com
Small programming tricks
blog.jim-nielsen.com
Can we stop with the uptime percentages?
autism.org.uk
Autism is genetic – and why this information matters now
x.com
📣 Speaker announcement @saranormous is the founder of the AI-native venture firm Conviction and has placed early bets on the most important companies of tomor…
news.ycombinator.com
Show HN: Give your AI agents access to WhatsApp
Hi, I built Chat-Man because I wanted a cheap way to give my agents access to WhatsApp without integrating a WhatsApp library separately in every project. You get WhatsApp MCP server, so you can connect WhatsApp to an agent and programmatically read, search, extract and send messages - but also have a web UI for non-techies. You can also receive webhooks for incoming messages for starred conversations. What you can do with the MCP? -Read WhatsApp messages and turn them into CRM records -Summarise conversations or groups -Extract structured information from messages -Find messages, participants or other WhatsApp data -Send messages -Manage group memberships Of course.. you always need to follow data protection regulation. The web UI: -Edit group details -See who joined or left -See who is most active -Export group members -Match members against a CSV (e.g. from a CRM) -Bulk message members -Bulk kick members -Create invite links For individual contacts, you can send messages directly from the UI. Why I built it for myself? I needed WhatsApp access for several projects (hermo.ai, florahaus.co.uk, …) and didn't want to pay for a WhatsApp integration for each project. Link: I’d love to get some feedback, especially from people building agents that need access to WhatsApp. Fabian
bbc.co.uk
Microsoft says AI rival Anthropic could have 'disastrous impact' on humanity
institute.deepmind.com
The DeepMind Institute
frvr.com
PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand"
agentlien.github.io
Anatomy of a Texture
I just uploaded a new article describing how a modern video game texture is stored in graphics memory, why it's done this way, and how it may differ between platforms. It is called Anatomy of a Texture.
mustafa-suleyman.ai
A warning about 'model welfare'
asteriskmag.substack.com
We know what a world without work looks like
openai.com
OpenAI expands ChatGPT ads with Sponsored Agents
arxiv.org
Dream-RSI: Recursive Self-Improvement through Evolving Worlds
wired.com
Hackers Got Inside a Flock Camera
oldmanrahul.com
I replaced my brown-noise browser tab with a menu bar app
cloudx.ai
Scaling Golang CI by Replacing actions/setup-go
stale.jock.pl
Show HN: How Stale Is Your AI? Release age and training cutoff for 20 models
github.com
ImpactGate: A merge gate that scores the structural decay AI adds
enclave.ai
DeepSeek v4.1 Flash Is Now Our Best Hacking Model
ycombinator.com
Kyber (YC W23) Is Hiring a Forward Deployed Engineer
tomshardware.com
Original Sony PlayStation 2 security chip 'broken wide open' after 26 years
manchestereveningnews.co.uk
Tiny village votes for 'independence' from UK amid asylum seeker row
gultsch.social
The Google Play app review process now regularly takes longer than a week
developer.nvidia.com
Nvidia announces native GPU programming in Rust
tomshardware.com
Devastated father says his 9-year-old son spent $118,000 on YouTube ads
status.salesforce.com
Salesforce Global Outage
techwontsave.us
Tech Fascism Has Come for American Democracy
bbc.com