Apple has begun rolling out its redesigned Siri AI with iOS 27, requiring a waitlist for access. Engadget's Mat Smith found the assistant much improved in testing.
Receipts: veriwire.news/daily-ai-news/…
An open-weights model matched Claude Opus 5 High on security research for $2.38 instead of $74.68 — Cantina's apex-flash-1 solved 40 of 60 held-out bug tasks, now free on Hugging Face under MIT.
1 story today, every one sourced → veriwire.news
Cantina Security and Yeta Labs released apex-flash-1, an open-weights vulnerability research model built as a GRPO fine-tune of Z.ai's GLM-5.3-Flash, available on Hugging Face under…
Receipts: veriwire.news/daily-ai-news/…
Starting in October 2026, Google will restrict free personal account users to its smallest Gemini model, Flash-Lite, while Flash and Pro require paid subscriptions.
Receipts: veriwire.news/daily-ai-news/…
Starting October 2026, Google's free users get only its weakest Gemini model, Flash-Lite — and $4.99/month subscribers lose Pro access.
NASA and IBM open-sourced a lunar foundation model trained on 17 years of orbiter data, cutting ice-prediction error up to 22 percent.
Aleph Alpha released Kolibri, an Apache 2.0 English-German MoE with 3.46B active parameters and over a million tokens of context.
A deep learning model spotted painful cats with 95.5% accuracy; ChatGPT, Gemini, Claude and Perplexity all scored it lower than a veterinary expert.
Google researchers built RRSI, which stops self-improving agents from memorizing tests, gaining 4.7 points on unseen benchmarks with about 30% fewer tokens.
6 stories today, every one sourced → veriwire.news
A smartphone-sized deep learning model developed by researchers at Université de Montréal distinguished painful from pain-free cats with 95.5% accuracy, using the Feline Grimace…
Receipts: veriwire.news/daily-ai-news/…
Meta is giving away 5,000 free Muse Home Link units to Muse subscribers, alongside open-source firmware and a Linux SDK for building Muse-connected gadgets.
1 story today, every one sourced → veriwire.news
Meta introduced Muse Gadgets, an open-source project letting developers build hardware that connects to its Muse AI agent.
Receipts: veriwire.news/daily-ai-news/…
NVIDIA will offer a 64GB configuration of the NVIDIA DGX Spark starting at $4,999 on Oct. 23 through Acer, ASUS, Dell, Gigabyte, HP and MSI.
Receipts: veriwire.news/daily-ai-news/…
OpenAI fired three safety researchers for allegedly leaking confidential information to an outside AI safety organization — and a fourth departed right after.
NVIDIA's 64GB DGX Spark arrives October 23 at $4,999, running models up to 100 billion parameters on device.
Amazon's cloud chief says over 100 data center moratoriums are being considered nationwide, and warned they could hurt the economy.
Shantou's pilot zone ships about 65 billion AI tokens daily to overseas customers while the computers stay in China.
A couple's GPT-4 role-play session suggested the MRI that found a 6-centimeter fibroid six years of fertility treatment missed.
8 stories today, every one sourced → veriwire.news
An MRI ordered at the insistence of Drew Housman and his wife revealed a 6-centimeter submucosal fibroid in her uterine wall, undetected during six years of fertility treatment.
Receipts: veriwire.news/daily-ai-news/…
Glow's PixelLeak report found that AI coding agents exposed over 13,000 internal screenshots from more than 300 organizations, including Fortune 500 companies and a frontier AI lab.
Receipts: veriwire.news/daily-ai-news/…
More than 13,000 internal screenshots from 300 organizations — including Fortune 500 firms — leaked because AI coding agents hosted them in public GitHub repos.
Grokipedia got a v0.3 redesign with a new logo and homepage, but its AI-cloned articles still haven't changed.
Photon raised $4.5M in seed funding to replace mobile apps with iMessage and WhatsApp agents, after signing 40,000 developers.
3 stories today, every one sourced → veriwire.news
Grokipedia, SpaceXAI's AI-powered Wikipedia competitor, received a v0.3 update with a new logo and refreshed homepage and live edits page, according to SpaceXAI head of design Benji…
Receipts: veriwire.news/daily-ai-news/…
Announcing Gemini 4 Argon, our new frontier model.
Argon is built to sustain deep reasoning across complex, long-horizon workflows and delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.
We’re also expanding the model’s output token limit to an industry-leading 1M tokens.
Argon is currently rolling out to a set of trusted cyber defenders in the Fairwind Program, with broader availability as soon as possible.
Introducing Gemini 4 Argon – our new frontier model.
It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
Google is paying roughly 100 digital publishers for content used in AI Overviews, AI Mode, and Gemini, The Information reports.
Receipts: veriwire.news/daily-ai-news/…
Zhipu's freely downloadable GLM-5.3 built working exploits in 50 of 410 ExploitBench attempts, nearly matching Anthropic's Claude Mythos Preview at 56.
Google is paying roughly 100 publishers for content used in AI answers — some under $1,000 — and the EU has opened an antitrust probe into the terms.
Sundar Pichai, Mark Zuckerberg, Elon Musk and other AI CEOs signed a voluntary safety accord with four rules and no penalties.
Sam Altman says OpenAI won't go public until it can make "confident safety claims," while rival Anthropic's IPO is expected in November.
Airbnb now lets users search rentals with text or voice prompts, with AI-generated filters and wish-list comparisons.
8 stories today, every one sourced → veriwire.news
101 Followers 145 FollowingBuilding specialist AI datasets • Fine-tuning & evaluating LLMs • ML student • Glassblower by trade, Sharing my workflows and the ups and downs of using agents
5K Followers 1K FollowingTesting local AI models on real tasks. Reviews, tools & practical workflows. Dev Ambassador @antlingagi @Alibaba_Qwen @openbmb
1K Followers 3K FollowingI help small businesses automate the boring stuff, customer replies, lead routing, CRM syncing using AI + https://t.co/bcJ4mZuGsn. DMs open
2K Followers 1K Following🚀 100 → 1k in 6 days
🛠️ Building apps with AI that save time
🎯 Next: 2K
👉 Follow for vibe coding, AI tools & X growth hacks
👇 Games, that might be cool
1K Followers 827 FollowingGrowth / @ https://t.co/uWAwlBbLF5 | Helping teams find the right prospects on LinkedIn and turn conversations into opportunities.
101 Followers 145 FollowingBuilding specialist AI datasets • Fine-tuning & evaluating LLMs • ML student • Glassblower by trade, Sharing my workflows and the ups and downs of using agents
1K Followers 137 Following@GeorgetownLaw '22.
Law firms → banking → now learning to build with AI.
Sharing what AI actually changes for lawyers and bankers.
3K Followers 2K FollowingBuilding https://t.co/Gp0xBjMvzS 🚀 https://t.co/Bo8bKF70Bq 💜 https://t.co/rYul4jV7uT 💬
Helping builders turn ideas into products. 20+ years making things for the web.
3K Followers 715 FollowingDay time artist and night time AI enthusiast. Building & benchmarking frontier LLMs on 4x DGX Spark clusters + Mac. Creator of Vesica Studio. Houston
787 Followers 2K Following(en)A kind of software engineer wannabe, living in Tokyo, Japan. Sometimes listening & playing music. most tweets written in Japanese... (jp)twitterは適当に更新します.
3K Followers 1K Following🚀 https://t.co/Kl1EcBHZU9 - Launch your startup on codehype and get discovered by google, llms & 100k+ users.
105K+ Followers on IG | Sharing what I learn ↓
409 Followers 405 FollowingOld grumpy IT guy pushing AI onto old hardware.
I test local AI so you don't have to.
Benchmarks, hacks, failures & lessons learned.