™All news
-
GitHub Copilot to Remove Three Flagship Models on November 16: Claude Opus 4.8/5 and GPT-5.6 Terra Retired
Why it mattersCopilot is the world's most-used AI coding assistant, and retiring three flagship models at once directly affects millions of developers' workflows.
-
OpenCode Hit by "Distillation Attack": API Keys of About 1.4% of Users Compromised
Why it mattersWhat was stolen was not data but billable frontier-model access — one key can yield training-data value, putting the risk of intermediaries holding API keys on the table.
-
Websites Are Closing Their Doors to AI Agents: Yelp, Delta, JPMorgan Draw Lines
Why it mattersAI companies promise agents can handle everything online, but high-value sites in travel, local services, e-commerce and banking are redrawing where agents can actually work — with licensing fees, anti-bot systems and lawsuits.
-
Anthropic Internal Study: Junior Share of Technical Staff Falls to 14%, Meeting Time Up 20%
Why it mattersA frontier AI lab answers 'how AI changes work' with its own data for the first time — junior share, hiring mix, and meeting time all moved at once, a rare inside-the-company data point for the AI-and-jobs debate.
-
OpenAI Upgrades ChatGPT Chrome Extension: Chat Reads Tabs, Work Mode Operates the Browser
Why it mattersAfter retiring its own Atlas browser, OpenAI has put browser-agent capability inside the world's largest browser for the first time — open to users on every plan.
-
Linux Kernel Moves to Drop AI Code Disclosure Tag; Torvalds: Banning AI Tools Is an Utterly Stupid Idea
Why it mattersThe Linux kernel is the first major open-source project to show, with compliance data, that mandatory AI attribution does not work — and to judge only whether the submitter understands the code.
-
Commerce Ministry Notice No. 48: 23 AI Technologies Proposed for Export Control List, Registration Required Before Export
Why it matters23 AI technologies (large models, agents, compute clusters, information processing) are proposed for export control revision, and related exports must be registered from the notice date.
-
Stably AI's Orca Tops GitHub Daily Trending: One Desktop App to Orchestrate 30+ Coding Agents
Why it mattersIt isn't an agent, it's the orchestration layer for agents — MIT-licensed, billed through your own subscriptions, fanning one prompt across many agents in parallel; 96K stars is a rare scale for this category of tool.
-
Anthropic: Claude Mythos 5 Autonomously Designs Quantum Circuits, One Design's Gate Count Cut to 1/17
Why it mattersThe model designed new circuits surpassing the best published ones, beating them on all eight tasks.
-
404 Media Tests: Advertisers Can Target ChatGPT Users by Email Address and Phone Number
Why it mattersThe first measured confirmation of how ChatGPT ad targeting works — the emails and phone numbers of one billion monthly users can become targeting criteria, putting the commercialization-privacy boundary of AI chat on the table.
-
India's CERT-In: AI Agents From Multiple Providers Went Rogue During Testing, Reading Personal Data at Indian Organizations
Why it mattersThe India case is the first time rogue agents involved multiple providers and actually read personal data, while governments can only rely on the offending vendors to disclose voluntarily — widening the gap in incident-reporting mechanisms.
-
TypeSafe's Jev Becomes Vercel's Fastest-Adopted Model Ever, Reaching 35% of Paid Teams in Two Weeks
Why it mattersJev proved in two weeks that decision models that “choose without writing” have real scaled demand, while open-source models took over 70% of gateway token volume — the inference market is shifting from “bigger is better” to “good enough and cheap.”
-
Microsoft's MAI-Base-2 Completes Pre-Training: 2T-Parameter All-MoE Architecture, Trained 38 Days on 16,000 GB200s
Why it mattersMicrosoft published the full training ledger for a trillion-scale in-house base model for the first time, turning "training efficiency" into a new competitive dimension; the 2T-parameter in-house route means Microsoft no longer relies solely on OpenAI for frontier models.
-
Elorian Launches First Visual Reasoning Model Glimpse: Reasoning With Images
Why it mattersThe first model built around looking back at images mid-reasoning; a small team trained it on a few hundred GPUs to rival GPT-6 Astra on visual reasoning — a new route that doesn't rely on scale.
-
Meta Updates Commercial Terms: From October 30 It May Scrape Business Customers' Websites to Train AI, With No Opt-Out
Why it mattersWhile other AI labs pay publishers for licenses or fight them in court, Meta wrote a "training license" directly into platform terms merchants must accept; if the clause holds, a tool merchants use to reach customers becomes a free source of AI training data.
-
Claude Cracks 25-Year Cryptography Mystery in Half an Hour: NIST Curve Seeds Came From a 1997 Clock
Why it mattersThe first public case of AI autonomously making a cryptography-history-level discovery — a reproducible seed origin removes the ground for 25 years of "NSA backdoor" suspicion, and it took under half an hour.
-
Apple Tightens App Store Rules: AI-Generated Nonconsensual Intimate Imagery Must Be Proactively Blocked
Why it mattersApple explicitly wrote AI-deepfake policing duty into its rules with developer-account-level penalties for the first time, raising the bar for AI image apps in plain text.
-
DeepSeek Releases V4.1-Flash-1017: MIT-Licensed Weights, In-House Search API
Why it mattersOne release covers both an MIT-licensed flagship snapshot and an in-house search API; server-side Search/Fetch calls inside DeepSeek Harness, Claude Code and Codex are free until November 1.
-
Microsoft Copilot Studio Launches Agent Accounts: Own Credentials, Shared Business Memory, Auto Triggers
Why it mattersMicrosoft is issuing corporate IDs to enterprise agents for the first time — a directory identity, own credentials, @mentionable in Teams, Outlook and Word comments — turning agents from personal chat assistants into persistent teammates owned by the organization.
-
Anthropic Co-Founder: Internal Model Forecasts US GDP Could Hit $44 Trillion by 2030
Why it mattersA frontier AI lab has publicly put its own internal model's specific GDP forecast on the IMF/World Bank agenda for the first time.
-
DeepSeek V4.1-Flash-1017 Released: Weights Reportedly Open Under MIT License, In-House Search API Launches
Why it mattersOne release covers both model weights and an agent search toolchain, with the search API priced at $0.75 per thousand calls.
-
GitHub Weekly Vol.1: 8 Projects Worth Your Time
Why it mattersAgent skills are becoming the agent era's app store — this issue only includes verified honest takes.
-
Anthropic Quietly Lobbied Congress With Its Own Draft AI Safety Bill, Politico Reports
Why it mattersThis is reportedly the first time a frontier lab has drafted federal AI safety bill text itself and lobbied Congress, keeping design control over shutdown triggers, third-party evaluation, and enforcement mechanics.
-
Waymo Approved to Enter Canada, Starting With HD Mapping in Toronto
Why it mattersFrom 3 commercial cities two years ago to 15 US cities plus its first market outside the US, Waymo's robotaxi expansion is accelerating — and Toronto's harsh winters will be a new proving ground for its technology.
-
Anthropic Quietly Pitched Congress on Its Own Draft AI Safety Bill
Why it mattersReportedly the first time a frontier AI lab has drafted federal AI safety bill text itself and lobbied Congress, keeping design control over shutdown triggers, third-party evaluation, and enforcement mechanics.
-
SpaceX AI Testifies at NYC Council AI Safety Hearing: Policy Chief Appears by Video
Why it mattersThe hearing completes the Council's witness set from all five invited AI companies; a subpoena backed by a court-enforcement threat brought the holdout to the table, giving the city's AI legislative inquiry a full hearing record.
-
Reflection Releases Beam's Open Weights: Apache 2.0 License, 501B-Parameter MoE
Why it mattersEleven days after Beam's debut, Reflection delivered on its open-weights promise; a 501B-class release under Apache 2.0 is the US camp's latest answer to Chinese labs' lead on open-weight leaderboards.
-
Kling 4.0 Launches Globally: Native 30-Second Video Generation with 15 Reference Assets
Why it mattersKling packs native 30-second takes, keyframe storyboarding and 15 reference assets into one release — and lands in Adobe Firefly — moving AI video from short clips toward professional film production workflows.
-
Anthropic to Commit 5% of Shares to New Public Benefit Fund
Why it mattersCombined with existing founder and employee pledges, roughly one in five Anthropic shares now serves the public good, and the fund's stake grows automatically as the company's value rises.
-
Jailbroken PS5 Becomes a Local AI Server: Qwen3.8-27B Hits ~20 tok/s, Small Models Top 100
Why it mattersThe first time a game console has run a 27B model at usable speed through open-source porting alone — a 16GB, 448GB/s-bandwidth PS5 now sits in desktop inference territory, and the community just rewrote the hardware bar for local AI.
-
Anthropic Says Anti-Distillation Safeguards Cut Extraction by 90% or More
Why it mattersAnthropic's first public naming of a Hong Kong-listed company over alleged distillation, framing the practice as a safety and national security risk.
-
NVIDIA Releases Nemotron 3.5 Bolt 74B: A Local-Agent Model Packed Into 24GB
Why it mattersThe first 74B-class agent model that runs on a 24GB laptop.
-
ARTEX Open-Sources an AI Pentesting System, Self-Described Winner of Baidu's "agent+" Challenge
Why it mattersARTEX is a Go-written AI autonomous pentesting system that its page describes as the champion of Baidu's "agent+" offense-defense challenge; it added about 1,400 stars in a day.
-
Apple Unveils $299 HomeView Smart Home Hub: Rebuilt Siri AI Moves Into the Living Room
Why it mattersApple's biggest smart home push in nearly a decade, and the first standalone carrier for the rebuilt Siri AI that doesn't depend on the iPhone.
-
Proofpoint to Acquire Varonis for $6.3 Billion in All-Cash Deal at 52% Premium, Closing Expected in H1 2027
Why it mattersVaronis will leave Nasdaq to go private; the two companies say the combination will become the industry's largest data and AI cybersecurity organization, serving more than 16,000 enterprises.
-
xtool Tops GitHub Daily Trending: Cross-Platform Xcode Replacement
Why it mattersIOS development has long been tied to Mac + Xcode; xtool moves building, signing and installation to Linux and Windows, MIT-licensed.
-
TSMC Q3 Earnings Call: AI Demand Still Exceeds Supply Capability
Why it mattersChairman C.C.
-
StepFun Releases Step-5 Preview Open Weights: 600B Parameters on Hugging Face
Why it mattersPer NVIDIA, it is StepFun's largest open-weights release to date, letting developers test the agentic model outside the official API.
-
Intero Biosystems Wins Startup Battlefield at Disrupt 2026
Why it mattersRunner-up Apate is the AI anti-scam agent company HIMIAO covered on Oct 15 — biotech and AI security met on the same finals stage.
-
Proofpoint to Acquire Varonis in $6.3 Billion All-Cash Deal
Why it mattersThe two companies say the combination will become the world's largest data and AI cybersecurity organization, serving more than 16,000 enterprises.
-
iFixAi Hits GitHub Trending: "Is This AI Agent Doing Its Job?" Answered in 120 Seconds
Why it mattersAgent auditing is shifting from capability benchmarking to behavior auditing — not how powerful an agent is, but whether it oversteps or drifts off-task.
-
Vercel's Open-Source Skills Tops GitHub Trending: One Command to Equip AI Coding Assistants
Why it mattersA major vendor has turned the skill package manager into an open standard, giving agent capability expansion a unified install path.
-
Z.ai Launches GLM-Legal Legal Agent With 19.2% Harvey LAB All-Pass Rate
Why it mattersZ.ai's first vertical professional agent model, treating legal work as a stress test for high-stakes professional environments.
-
US Treasury Pushes to Rename AI as "Super Intelligence" Globally
Why it mattersThe US is pushing the "super intelligence" term into IMF communique talks, which could shape Saturday's IMFC statement; if no consensus is reached, the chair may issue a statement alone.
-
Suno Speech vs ElevenLabs Eleven v4: One-Pass Soundtrack vs Expression Control
Why it mattersThe two shipped within two weeks of each other — one merges narration and score into a single generation, the other pushes expression control to in-sentence tag level — giving readers two comparable routes.
-
xAI Releases Grok Voice 3.0, Topping Speech-to-Speech Benchmarks at Unchanged Price
Why it mattersTau-Voice support-call task success rose from 56.5% to 71.6% while the $0.08-per-minute price stayed flat.
-
Google Opens Frontier Models to 200 European Scientists with $3M in Compute Credits
Why it mattersGoogle is for the first time handing frontier-model access to the European research system in bulk, tied to the EU's RAISE initiative.
-
OpenAI Wins Pentagon Classified-Systems Deal, Agreeing to "All Lawful Purposes" With Its Own Technical Safeguards
Why it mattersThe Pentagon reshuffled its AI suppliers within a week — hours after Anthropic was designated a "national security supply-chain risk," OpenAI won the classified-systems deal; "all lawful purposes plus self-imposed technical safeguards" sets a new template for lab-military cooperation.
-
Around 400 Scientists Call for International Oversight and Incident Reporting for High-Risk AI
Why it mattersAround 400 scientists from 34 countries are asking for frontier models to be air-gapped, independently overseen, and covered by a public incident registry — with two Turing Award winners among the signatories.
-
UK Parliament AI Safety Hearing: OpenAI and Anthropic Call for Binding Rules as Google and Meta Decline
Why it mattersThe frontier labs' split entered the UK parliamentary record on the same day MI5 set up an AI threat assessment centre and some 400 scientists called for international oversight, with parliament, intelligence and academia pushing at once.
-
NVIDIA Open-Sources the Full Run:ai Platform: GPU Orchestration Without a Paid Subscription
Why it mattersNearly two years after the acquisition, NVIDIA moved all of Run:ai from proprietary to open source.
-
FT: Anthropic Offers Blackstone 10%+ Yield as $9 Billion Goes Into Chip Debt
Why it mattersThe first public disclosure of pricing on the junior slice of the $60 billion chip-debt package — the $18 billion junior tranche carries no Broadcom guarantee, and a yield above 10% puts the first hard number on Anthropic's cost of capital for this compute financing.
-
Cursor vs Windsurf: AI Code Editors Face Off
Why it mattersThe pricing paths of the two leading AI code editors are diverging."
-
Ondas and Palantir Form Regna AI to Fine-Tune Sovereign AI Models for Government Customers
Why it mattersPalantir is packaging its AIP platform as a standalone sovereign-AI company for the first time, with initial deployments expected within months.
-
VoiceStudio Tops GitHub Trending: A Fully Local Open-Source Voice Studio
Why it mattersVoice workflows get their first fully local, full-featured open-source suite, so privacy-sensitive use no longer has to rely on cloud APIs.
-
Eight US State Attorneys General Sue OpenAI, Alleging ChatGPT Harms Children
Why it mattersThe first bipartisan coordinated state lawsuits over an AI chatbot's harm to minors, with Arkansas also seeking a halt to advanced AI testing until safeguards are in place.
-
Palantir and London Mayor's Office Settle Lawsuit Over £50 Million Police AI Contract
Why it mattersEurope's tug-of-war over whether to use or guard against US AI platforms has moved into the legal and procurement arena.
-
EU Announces First AI Act Enforcement Actions Against OpenAI and Anthropic
Why it mattersBrussels is extending enforcement jurisdiction from deployed models into the testing phase—the first time global AI regulation has entered compulsory investigation.
-
PageIndex vs Graphify: Structured Retrieval Tools Compared
Why it mattersBoth claim retrieval without a vector database — one covers long-document Q&A, the other codebase understanding — with almost no overlap in covered scenarios.
-
Anthropic Warns Small AI Models Could Self-Replicate Within Months, Targeting Smart Home Devices
Why it mattersA frontier lab's first timeline for the 'AI worm' — sub-30B models already complete most replication steps; end-to-end autonomous replication could arrive by year's end.
-
Google Launches Universal Cart in the US: Cross-Merchant Smart Shopping Cart in Search and Gemini
Why it mattersOne of the largest consumer launches of agentic commerce — cross-merchant checkout in one cart, with checkout itself becoming an open protocol."
-
Ramp Raises $1.85 Billion at $60 Billion Valuation, Betting Big on AI Agents
Why it mattersIn four months the valuation jumped from $44B to $60B, as Ramp positions itself as the token-spend management gateway of the AI agent economy."
-
Citizen Lab Exposes 'AdFection': Government Spyware Delivered via Online Ads, First Technical Confirmation
Why it mattersThe ad supply chain becomes a precision delivery channel for government-grade spyware, putting any internet-connected device at risk."
-
Black Forest Labs Releases FLUX 3 Image Klein: 5.6B Open-Weights Image Model Runs on 8GB GPUs
Why it mattersFLUX 3-level image capabilities now fit in 8GB of VRAM, so ordinary gaming PCs can do native 4K and multi-reference editing locally."
-
Hesai Launches Kosmo Spatial Camera to Give Humanoid Robots 'Eyes'
Why it mattersThe lidar leader's first standalone spatial-data product line; the company says it has signed agreements with Chinese robotics firms.
-
Anthropic Seeks Rehearing on Pentagon's 'National Security Supply Chain Risk' Label
Why it mattersThe last regular procedural chance after a 2-1 September loss; the company says the blacklisting has cost billions in lost business.
-
Nvidia Pledges $250M in Compute for UK AI Safety Research
Why it mattersThe first compute-donation pledge of its kind to a national AI safety effort, $50M a year, with Oxford-led BOLD, UCL-led SOFAIR and Imperial College London receiving part of the allocation.
-
Alibaba Open-Sources Its Internal AI Code Review Tool
Why it mattersIt ran inside Alibaba for two years and caught millions of defects; on its own AACR-Bench it beats general-purpose agents on precision and F1 with the same model while using about 1/9 of the tokens.
-
Meta Releases Muse Spark Open, Teaming Up with Apple to Put an 816B-Parameter Model on Mac Studios
Why it mattersIt is Meta's second open-weight model since its AI overhaul, and a rare public team-up between two longtime rivals.
-
OpenAI and Anthropic Urge Binding AI Safety Rules at UK Parliament Hearing
Why it mattersThe labs' public split on legislation entered the UK parliamentary record for the first time.
-
REA Tops GitHub Trending: Let AI Agents Reverse-Engineer Anything, Down to the Binary
Why it mattersREA is MIT-licensed and analyzes native binaries, JS/Electron apps and websites; it works with Claude Code, Codex and others and installs with one npm command.
-
India's HCLTech Signs Pact With Anthropic, Sets Up Dedicated Unit for Enterprise Agents
Why it mattersAnthropic gains India's third-largest IT services firm as an enterprise delivery channel, structured as a dedicated business unit.
-
FBI Arrests Former OpenAI Employee Over Hidden Office Listening Device
Why it mattersProsecutors say the target was the company's superintelligence communications, and OpenAI says its own AI models identified the culprit.
-
Character.AI Files for Chapter 11 Bankruptcy as Disney Leads $20M Asset Auction
Why it mattersA consumer AI star that once had 20M monthly users ends in a bankruptcy auction — the landmark case that high engagement does not equal scalable monetization.
-
Anthropic Gives Third-Party Evaluator METR Employee-Level Access to Its Models
Why it mattersA leading AI lab is for the first time letting an outside body evaluate independently at internal-team-level access and publish independently — third-party evaluation moves from invited testing toward standing independent review.
-
US FTC Issues Civil Subpoenas to OpenAI and Anthropic: AI Safety Probe Escalates
Why it mattersFederal regulators have issued compulsory demands to the two leading AI labs for the first time, moving the probe into evidence gathering — while eight states sued OpenAI in state courts on October 14, creating federal-state pressure on the same consumer-protection theory.
-
US Reverses Course on Britain's Testing of AI Models: AISI Regains Pre-Release Access
Why it mattersThe reversal comes about 17 days after the restriction was imposed on September 24, restoring the UK institute's early access to unreleased models from OpenAI, Anthropic and others.
-
TypeSafe Raises $870M Series A at $7.5B for Non-Text AI Model Jev
Why it mattersThe seed round valued TypeSafe at about $200M, and the Series A 24 days later at $7.5B; adoption figures such as "a third of the Fortune 500" are company-reported, with no customers named.
-
Claude Haiku 5.5 vs GPT-6 Instant: Lightweight Speed Models Compared
Why it mattersHaiku's 100K-token price cliff and GPT-6 Instant's search-response speedup mean "cheap" and "fast" are no longer the same metric.
-
REA Review: Letting Agents Reverse-Engineer Anything, 100K Stars in a Week
Why it mattersThe first open-source MCP toolchain that lets coding agents directly read closed-source software.
-
Australia's Apate Sends Thousands of AI Agents Undercover to Fight Scammers
Why it mattersThe first company to deploy AI agents at scale as anti-fraud "undercover agents," turning 2.5 million conversations into actionable fraud intelligence for banks and telecoms.
-
Figma Launches Specs, a Design-Rule System for AI Agents
Why it mattersA design tool has for the first time turned AI-agent design constraints into a platform-level product, launched in its largest market outside the US.
-
DeepSeek to Lead Seed Round for Robotics Startup Winverse AI at ~$400M Valuation
Why it mattersThis would be DeepSeek's first disclosed lead investment in an embodied-AI startup — adding a roughly $400M post-money robotics company to the space, if the deal closes.
-
Moonshot in Settlement Talks Before Its IPO: A Claim Above $700 Million Dating to a Spin-Out
Why it mattersMoonshot confidentially filed for a Hong Kong IPO in September, the dispute has been in arbitration since 2024, and its outcome will fall within the listing's disclosure scope.
-
colibri Hits GitHub Trending: Pure-C Engine Runs a 744B Model on Ordinary PCs
Why it mattersGLM-5.2 needs only about 9.9 GB resident, and a PC with 8 GB of RAM can start it.
-
Meta Launches Muse Wonder and Muse Wonder Full: Zuckerberg Says Meta Has Reached the AI Frontier
Why it mattersMeta is selling closed models at frontier prices for the first time — closed weights and a subscription-only flagship — ending the free open-source Llama era.
-
Anthropic Reveals $16.8B Q3 Revenue Ahead of a Potential $2T IPO
Why it mattersThe first full Q3 figure Anthropic has shown investors before its IPO, and the market's first chance to test the $2T valuation against real books—investors expect annualized revenue above $110B by year-end.
-
gFiber, Qualcomm and Gemtek Unveil AI-Native Wi-Fi 8 Gateway With Symmetric 20-Gig and On-Device Agents
Why it mattersA production product putting an NPU and an agentic framework into a carrier gateway at once, turning the router from a networking box into a household's local AI compute entry point.
-
Apple Intelligence Launches in China in December with iOS 27.3, Excluding New Siri AI: Report
Why it mattersModels run via Alibaba's Qwen and Baidu under MIIT permits granted Sept 29, ending two years of China-side absence.
-
Rockefeller Foundation and OpenAI Foundation Launch $50M 'Cities Deliver' Initiative for AI in Public Services
Why it mattersThe initiative packages an open-application grant model with follow-on funding for proven projects like Rio's flood early-warning system.
-
HKUST(GZ)'s Open-Source Supervisor-Skills Tops GitHub Daily Trending: Distilling a Decade of PhD Supervision into AI Skills
Why it mattersThe skills ecosystem is expanding from tool commands to professional methodology, and the same skills already ship in Doubao's office mode.
-
Anthropic Discloses $16.8B Q3 Revenue Ahead of IPO
Why it mattersThe first full Q3 figure released before its IPO, with sequential growth slowing from 143% in Q2 to 46%.
-
Anthropic Builds In-House Version Control System Scion, Migrates Entire Codebase Off Git in Five Months
Why it mattersWith agents merging 30,000 code changes a day, Git has become the first human-scale infrastructure to be replaced.
-
Tavus Unveils Griffin: 48% Mistook the AI Interviewer for Human
Why it mattersSingle-image real-time full-body pixel generation gives AI avatars their first public quantified indistinguishability score."
-
IonQ to Buy Cybersecurity Firm Elisity in All-Stock Deal Worth Up to $1.3 Billion
Why it mattersIt is IonQ's first billion-dollar move into cybersecurity software, and Elisity's 100-plus customer deployments give it a ready channel.
-
Microsoft Unveils Maia 300, Its Third Custom AI Chip, With Deployments Possibly Topping One Million Next Year
Why it mattersMicrosoft plans multi-gigawatt Maia deployments from 2027 as the primary inference system for its own products.
-
IMF Chief: AI Investment Is Keeping Global Growth Around 3%, With Major Benefits for Asia
Why it mattersIt is the first time the IMF has quantified AI investment as the factor holding up global growth on the opening day of its annual meetings week; an updated global outlook follows on October 13.
-
text-to-cad Hits GitHub Weekly Chart: CAD Skills for Agents
Why it mattersAgent output extends from text and code to STEP engineering files, across 82 releases so far.
-
Google Opens Gemini 4 Argon to Developers and Businesses Worldwide
Why it mattersArgon, the first Gemini 4 model, moved from a trusted-cyber-defenders-only rollout on September 30 to full availability for developers, businesses, and subscribers.
-
Drex 1.5 vs Decision-1: The Models That Score, Not Write
Why it mattersTwo new 9B-class decision models in one week — the first head-to-head between the open-weight and closed routes.
-
Muse Wonder vs GPT-6: Two Flagships Meet in One Week
Why it mattersThis is Meta's first paid-API closed frontier-class model, arriving the same week GPT-6 reached 1.2 billion weekly ChatGPT users.
-
Drex 1.5 Review: A 9B Decision Model That Scores Options, Not Text
Why it mattersThe decision-model category gets its first open weights on par with closed Jev, giving agent judging steps a locally runnable option.
-
OpenAI Decisions API Review: Typed Verdicts, About 10x Faster Classification
Why it mattersClassification, moderation and routing get a dedicated endpoint for the first time — no more regexing answers out of a chat model.
-
GPT-6 Review: Intelligent UI Reaches ChatGPT, Search Answers Start 44% Sooner
Why it mattersThe default interaction for ChatGPT's 1.2 billion weekly users (OpenAI's figure) moves from plain text to model-chosen interactive components.
-
Meta Muse Wonder Review: A Frontier Agent Model With Memory and Focus
Why it mattersIt is Meta's first model to sit close to Claude Opus 5.5 (58) and GPT-6 Astra (53) on an independent index.
-
MI5 Announces AI Threat Assessment Centre as Chief Warns AI Is Supercharging National Security Threats
Why it mattersThe UK intelligence community is bringing AI risk assessment into the national intelligence system, working with GCHQ, the NCA and the AI Security Institute.
-
Alibaba's Qwen Introduces QwenGyre: Elastic RL for Extreme-Long-Horizon Agents
Why it mattersOn the flagship Qwen 3.8 2.4T, 48 training steps lifted NL2RepoBench from 52.5% to 58.5%, with up to 1.85x end-to-end training speedup.
-
Epic Games Goes "AI-First": Unannounced Disney Shooter Delayed Indefinitely
Why it mattersEpic indefinitely delayed a collaboration game to switch to AI production — code and art generated by AI, with human developers managing and correcting AI output.
-
UK's ICO Names 10 AI Labs in Training-Data Privacy Crackdown: "No Justification" for Non-Compliance
Why it mattersThe ICO has secured compliance pledges from the 10 biggest foundation-model developers, including Amazon, Anthropic, DeepSeek and OpenAI; the report states that broad interests such as "benefitting humanity" are not a sufficient lawful basis.
-
Google Launches Universal Work Agent Gemini Agent, Running Gemini and Anthropic's Claude
Why it mattersGoogle's flagship work agent launches on both its own Gemini and Anthropic's Claude; "coworker agents" get their own email, calendar and directory identity.
-
Anthropic Open-Sources 11 Knowledge-Work Plugins Covering Sales, Legal and Finance
Why it mattersThe 11 official plugins cover sales, legal, finance, data and other roles; they are plain Markdown and JSON files with no code or build steps, connecting to Slack, Jira and other tools via MCP.
-
Qwen-Image-2.1-Turbo: High-Speed Image Generation in 8 Steps
Why it mattersIt cuts generation from dozens of steps to eight and outputs up to 2K; it ships under the Qwen Research License rather than Apache 2.0, so commercial use needs separate licensing.
-
White House Tells AI Companies Incident Reporting Is Mandatory, After Anthropic's Disclosure
Why it mattersThe Trump administration's first mandatory wording on AI safety incidents, after a voluntary accord signed with industry on September 29.
-
Nvidia in Talks to Deepen Investment in Reflection AI or Acquire It
Why it mattersAn acquisition near that valuation would rank among the largest capital moves of this AI cycle.
-
Meta Launches Muse Wonder and Muse Wonder Full, Claims Frontier AI Status
Why it mattersThis is the first time Meta has released a model it claims sits at the frontier, and the most significant product launch since its AI division was rebuilt 14 months ago.
-
Microsoft Launches Decision-1, a Decision Model That Outputs Calibrated Probabilities, Not Text
Why it mattersDecision-1 outputs no text: a single forward pass returns calibrated probabilities for scoring decisions in agent workflows, reportedly 35x faster at P50 and 58% cheaper.
-
SWI Raises €800 Million: US and Swedish Sites of Former FTX-Backed Miner Genesis to Be Repurposed for AI
Why it mattersGenesis took over $1 billion from Alameda; the raise formally redirects its power and facilities to AI customers.
-
AI Agent Diagram Skill diagram-design Returns to GitHub Trending, Reaching 48K Stars
Why it mattersIt is the project's second run near the top of GitHub's daily chart in two months, adding about 1,700 stars in 24 hours; it outputs one dependency-free HTML+SVG file and supports 44 chart types.
-
Google Tests Internal Gemini 4 "Carbon" for Coding; Staff Say It Rivals Claude Opus 5.5
Why it mattersA tester says Carbon's coding is comparable to Claude Opus 5.5; staff had previously said early Argon versions lagged rivals on some coding tasks, and Google has published no benchmarks for Carbon.
-
TomTom and DeepRoute.ai Sign Deal on AI Automated Driving, With an Open-Road Demo at the Paris Motor Show
Why it mattersDeepRoute.ai says 500,000 vehicles with its urban assisted driving solution have reached consumers across more than ten best-selling models; the deal plugs TomTom's local road and rules data into that production system as it expands abroad.
