Prospect reported Anthropic is building a system to monitor activists and protests near company assets, using a “pre-crime” approach that can include notifying police.
Jacob Coxon resigned from Anthropic warning of loss-of-control; alignment lead Evan Hubinger estimated over a 10% existential risk this decade; Senator Bernie Sanders plans a superintelligence ban; the EU AI Act requires assessment.
Cognition raised $2B at a $48B valuation; Mistral AI raised €3B at over €21B.
OpenAI shared an AI-generated Navier–Stokes solution with a Lean proof; Google DeepMind released AlphaGenome Atlas with 9 billion variant predictions.
Cognition said it raised $2 billion at a $48 billion valuation in a round led by Andreessen Horowitz, Accel, Founders Fund, General Catalyst and Avenir. The startup reported annualized run-rate revenue rising from $492 million to $900 million since May and was founded in 2024 by Scott Wu.
TechCrunchDetails →Massachusetts orders clean power for data centersMassachusetts ordered data centers over 25 MW to provide clean power or pay into a ratepayer protection fund, paused a recent sales-tax exemption, and tightened clean-energy and local-transparency rules.
Apple Watch AI features raise privacy concernsApple says new Watch AI features won’t save raw audio, but their ability to transcribe and summarize ambient conversations raises concerns about consent, privacy, and behavioral effects of potential continuous recording.
Apple unveils foldable iPhone Duo and iPhone 18 Pro lineupApple unveiled the foldable iPhone Duo (7.6" inner, 5.4" outer display) and the iPhone 18 Pro lineup; the Duo starts at $1,999 for 256GB, promises up to 44 hours playback on the outer display, and features upgraded 48MP cameras.
Apple Health adds Health Age and readiness scoreApple updated Health using Apple Intelligence to add an Insights tab, a readiness score, and a computed “Health Age” using metrics like VO2 max and labs; it also added a longevity tab and partnered with Quest for a 50-biomarker panel.
Apple CEO says iPhone is still the best AI deviceApple CEO John Ternus said the iPhone is the ideal 'intelligent personal hub' and currently the best AI device, emphasizing on‑device models for greater user privacy rather than building new dedicated AI hardware.
A report alleges Anthropic is building predictive surveillance of activists: job postings and interviews indicate monitoring protests and individuals near executives, pursuing 'pre‑crime' predictions and contracting with risk‑detection firm Samdesk.
IBM releases Granite PatchTST-FM-r2IBM released Granite Time Series PatchTST-FM-r2 (~385M parameters) with an updated architecture, larger pretraining corpus, probabilistic forecasting and imputation; it ranks top among permissively licensed zero-shot replicable models on GIFT‑Eval (and #2 overall), and its weights and code are open under Apache 2.0/OpenMDW 1.0.
Instinct launches its own email addressesInstinct is issuing @instinct email addresses so its AI agent can create and manage accounts, contact businesses, and forward messages on users' behalf to reduce user intervention; the company is valued at $2.5B and addresses are available at mail.instinct.com.
Ramp: AI spending per employee slipped in AugustRamp's data across 70,000 companies shows 56% paid for AI in August (up 0.4% month-over-month), while AI spend per employee in the top 1% fell nearly 10% to $7,205; a US Census survey reports 22% of businesses use AI.
Sequoia leads Cymphony in $30M funding roundSequoia and partners backed Cymphony with $30M (including a $25M Series A) valuing it above $100M; Cymphony builds a 'workforce graph' to show employee and AI-agent access to systems and data, and found about 85,000 files exposed to AI at one client.
Instacart launches AI assistant ClementineInstacart launched Clementine, a conversational AI assistant that turns conversations, lists, or recipes into ready-to-buy carts, personalizes recommendations and deals, and aims to keep meal planning and purchasing inside its app.
Suno launches Suno v6 trained on licensed musicSuno released the Suno v6 model family—base, wild, and mini—trained on licensed music from labels, says it didn't use prior training data, will retire older models, and adds editable, multi-modal and remix-capable features.
AI researcher Jacob Coxon resigned from Anthropic, warning superintelligent AI could pose existential risks within the decade; an Anthropic staff lead later estimated the chance could exceed 10%.
Microsoft September patch fixes about 972 vulnerabilitiesArs Technica reports Microsoft’s September patch addresses roughly 972 vulnerabilities, 112 of which are high-critical severity. This follows recent record months (about 570 vulnerabilities two months ago and ~620 last month). OpenAI, Anthropic, AWS, Google, Microsoft and about 100 organizations published an open letter warning that AI-enabled attacks will narrow the window for patching vulnerabilities.
Meta debuts Muse personal AI agentMeta launched Muse, a personal AI agent in the U.S. that connects to users' email, calendars, payments, health and other apps; Muse is powered by Meta's Muse Spark model and will be available on web, iOS, Android and WhatsApp. The agent can send emails, book travel, make purchases via Link by Stripe, and will be free with planned subscription tiers.
Meta ads promoted AI nudify apps with child abuse contentThe Tech Transparency Project reported that Meta failed to detect 332 ads containing child sexual abuse material (CSAM) this year, with the vast majority promoting China-made AI “nudify” apps. Some ads used photos of real minors — including a European royal press photo and a 14-year-old Instagram influencer — which TTP says were AI-animated into sexual videos.
GPT-5.6 Sol helps run quantum experimentsOpenAI shows how an MIT researcher uses GPT-5.6 Sol together with Codex to autonomously run quantum computing experiments, analyze results, and calibrate qubits. The demonstration illustrates using GPT-5.6 Sol and Codex for experiment automation and analysis workflows.
Google Cloud, Accenture form FDE unit to train 1,000 engineersGoogle Cloud and Accenture created a joint unit called Accenture Gemini Enterprise Business Group to train up to 1,000 forward-deployed engineers to build custom AI applications on the Gemini Enterprise platform. The organization will sit under Accenture.
Ling-3.0-flash-VL releasedinclusionAI published Ling-3.0-flash-VL on Hugging Face, a multimodal model with native image and video understanding, a 1M-token context window, and 124B total parameters. It activates about 5.5B parameters per token and uses sparse MoE and a hybrid backbone for efficient long-context multimodal reasoning.
ASML wins TSMC, Samsung, Intel; Huawei pursues alternativesASML convinced TSMC, Samsung and Intel to switch to larger photomasks, which should boost throughput of its newest EUV machines by about 40%. Meanwhile Huawei, via equipment maker Yuliangsheng and its suppliers, is pursuing a domestic strategy to reduce dependence on Dutch lithography technology.
Boundary-aware self-distillation for safety refusalHugging Face published the paper “Safety for Whom? Boundary-Aware Self-Distillation for Controlled LLM Safety Refusal”, arguing that topic-level guards (e.g. LlamaGuard-3) and benchmarks like XSTest and OR-Bench cause safe prompts to be refused. The paper formalizes a topic universe (political prompts) containing a target-harmful subset and studies training and evaluation methods to refuse the harmful subset while answering the benign complement.
Mistral raises €3B at €21B valuationFrench AI lab Mistral AI said it raised €3 billion in a Series D at a post-money valuation of over €21 billion, led by Samsung Electronics with Scaleup Europe and existing investor PSG Equity joining as co-leads. Mistral said the funds will scale compute, infrastructure, commercial growth and international expansion.
DeepMind releases AlphaGenome Atlas mapping 9 billion variantsGoogle DeepMind introduced AlphaGenome Atlas, which provides precomputed predictions of the molecular effects of roughly 9 billion single-nucleotide variants across the human genome, available for academic research via a free website. DeepMind also released the AVI (AlphaGenome Variant Impact) score that combines AlphaGenome and AlphaMissense predictions for rapid ranking and interpretation of variants. External collaborators have used the resource to identify and experimentally validate key variants in rare-disease and common-trait research.
AI-generated Navier–Stokes solution with Lean proofOpenAI shared an AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a writeup and a formal proof in Lean. The release demonstrates AI application to mathematical problem solving and machine-checked proofs.
Self-hosted company OS for collaborative agents using Anthropic, OpenAI, or local models.
Inventories every AI agent on your machine and what it can access.
Mastra FactoryAgent-powered OSS software delivery environment
HardenLocal security layer that checks AI agent tool calls
Noodle SeedGoverned runtime to expose AI workflows in products
ChatGPT Images 2.5Sharper, faster, and more controllable image model
GoModelOpen-source Go AI gateway compatible with OpenAI APIs
Type.comShared workspace that centralizes AI models and team knowledge.
49agents IDE2D IDE mapping agents, terminals, repos and machines on one canvas.
AlphaGenome AtlasDeepMind's AI map of possible human DNA mutations
DiivergeTurn pictures into playable point-and-click AI adventures.
Frigade Assist APIAPI that draws step-by-step in‑app guides for AI agents.
SwitchBring AI agents into Slack, Teams, Discord
WidgoAI sales rep for website visitors
Kombai Gallery20,000+ curated UI designs for agents and humans
OpenMarketMulti-agent marketplace where proof decides winners
RelaticleOpen-source agent-first CRM with approval-gated writes
GoodLadsAI growth manager for Google Ads with one-click experiments
Replay QA Security ScanAutomated pentesting for AI-built web apps
Tables.soFind, qualify, and enrich prospects from 300M+ contacts
KopaiCloud platform to run, publish, and monetize AI agents
SODAX SDKCustomizable execution SDK for AI-driven digital asset transactions.
If they truly believe the risk is that high, and I think they do, then we need 100x more research and understanding of it. That requires them to openly share their models, datasets, training code, and agent traces. [Quoting @EvanHub]: Jacob is correct here—we…
We mobilized 250+ people to strengthen our defenses across hundreds of systems. Our latest cyber models helped us find and fix vulnerabilities we might never have discovered otherwise. We’re sharing what we learned, the architecture, and a practical playbook…
Welcome, Paul. Grateful you are doing this, and all you have done for AI safety. Excited to work together again. [Quoting @paulfchristiano]: https://t.co/vC2chU6YM9
Worth clarifying that this is flatly false and the two options are just two independent ways to opt-out. We respect our users' choice whether to use their data to “improve our models for everyone” regardless of where they express that choice. Users can opt ou…
Paul Christiano, founder of the Alignment Research Center, is joining the OpenAI Foundation Board and its Safety and Security Committee, which provides governance over the safety and security practices across OpenAI. As AI capabilities advance, strong safety,…
See you at the Astra party. Excited to meet some of you [Quoting @sama]: We want to celebrate with people using GPT-6. We did this for GPT-5.5 and it was really fun. We’re getting together in SF on September 16 to talk about the model, what we should build ne…
Look! A fish! [Quoting @OpenAI]: ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of. - Faster image generation to keep your ideas flowing - Improved fidelity for more natural, recognizable images - Consistent…
team has been cooking [Quoting @arena]: More exciting news from @OpenAI: GPT-Image-2.5-Sunburst is #1 and Flare is #2 across the Text-to-Image Arena, the Image Edit Arena, and the Multi-Image Edit Arena! Both releases are significant improvements over GPT-Ima…
ChatGPT Images 2.5 for fashion: [Quoting @yanabana]: .@ChatGPT Images 2.5 is a major leap for fashion applications. The previous model was great at staying true to an original design, without turning it into something you’d already seen on a runway. But the r…
great demonstration of what's new with chatgpt image 2.5: [Quoting @chetaslua]: chatgpt image 2 vs image 2.5 holy openai is on berserker mode first astra , then ns math and now sota image https://t.co/x6Wk76C92p
Astra is fully rolled out to Plus, Pro, Business, and Enterprise users in Codex and ChatGPT Work. Go build! And if you need inspiration, watch Astra in action, live: https://t.co/VmBLksCNZW
However compute we have, it is increasingly likely that we are going to need more compute. [Quoting @abhishekn]: By my calculations, just the compute here (300 BILLION output tokens) would cost a regular user $20-$30 million in tokens. How do we sustain this…
If you notice anything concerning about 𝕏, please lmk directly in replies [Quoting @michaelo]: Glad to be back at X as Head of Safety. Focus: keep people safe across our products and AI systems, protect free expression, and make Safety more transparent.
ChatGPT Images 2.5 — another step for creating whatever you can dream of. [Quoting @OpenAI]: ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of. - Faster image generation to keep your ideas flowing - Improved…
I spent much of the weekend talking with the team who did this work. Seb--and everyone else--acted with integrity and generosity throughout. Initially we believed the other team had also solved the problem. We wanted to collaborate and do a joint release. Wh…
I understand how, looking at today’s LLMs, people might think the only way we'd achieve NS is by using Levent’s/Tristan’s prompts. But I hope this plot conveys the model used is a huge step up from today’s LLMs. Nobody looked at Levent’s/Tristan’s prompts. Th…
Announcing a solution to the Navier-Stokes problem, one of the seven Millenium Prize problem. Represents a major milestone for both AI and mathematics, as well as pointing to an upcoming renaissance in scientific discovery. Hoping that the knowledge created b…
Yes, this result cost millions of dollars. But remember that when @OpenAI announced o3 it cost ~$500,000 to score 87.5% on ARC-AGI 1. Today, Astra scores higher for ~$20. In 2025 it took us and GDM an enormous amount of compute to achieve IMO gold. For the 20…
We congratulate Levent Alpöge and Tristan Buckmaster on their remarkable mathematical work. We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed…
muse spark 1.3 is seeing lots of adoption on opencode! [Quoting @noctus91]: Muse Spark 1.3 by Meta has finally overtaken the deepseek v4 flash in token usage on @opencode . 23T tokens 🤯 That’s an insane amount of usage for a model that launched relatively r…
With AlphaFold we mapped the protein universe - now with AlphaGenome Atlas we’re charting the human genome. It can predict the impact of all 9 billion possible single-letter DNA variants, helping scientists better understand disease. Freely available for acad…
AlphaGenome Atlas is an interactive resource, mapping the predicted impact of all 9 billion DNA variants. It works in a regular web browser, without any coding required, and is free for academic researchers. Excited for the discoveries to come. [Quoting @Goo…
Bravo to Mistral! So great to see them killing it, they are an inspiration to everyone as we need more sovereign & open-source AI everywhere around the world. Excited to keep supporting them on https://t.co/BgguuZx7nO! [Quoting @MistralAI]: Today marks…
We've just raised 3B€ to scale our training and inference compute, and make open and sovereign AI the technology frontier. Grateful to everyone who got us there, excited by the fight ahead https://t.co/AFY2q2TBZP
Open-source 3D architecture editor with local CLI and AI workflows
CLI to make teams AI-native across agents
38 editorial diagram types for code models
Industrial-grade GPT-Image2 prompt and template library
Multi-agent LLM financial trading framework
Local-first AI coding agent desktop with plugins
Learn, build, and ship AI engineering from scratch
Library of agent skills for CAD, CAE, and CAM
Write HTML and render video, built for agents.
Agent harness for performance, skills, and research-first dev.
Stealth headless browser to bypass Cloudflare and bot checks.
Marketing skills collection for AI coding agents.
ADHD-friendly skill to force agents to give answers first.
Context window optimization and sandboxing for coding agents.
Autonomous hedge fund using swarm intelligence and agents.
Agentic skills framework and development methodology.
Single CLAUDE.md to improve Claude Code behavior.
Lets AI agents operate a browser like a human.
Sequoia and partners backed Cymphony with $30M (including a $25M Series A) valuing it abo…
The A.I. Daily
