AI News

Global artificial intelligence news: model releases, research, companies, policy, safety and the developments shaping AI worldwide. AIMag is based in Norway and publishes in English and Norwegian.

Here we collect everything we have published under AI News: news, analysis and explainers from the world of AI. The page is updated whenever a new story is published.

AI News

Latest stories

Alibaba Swaps Away Apache 2.0: Qwen-Image-2.1 Is Free to Download, but No Longer Free for Commercial Use

Sol costs half as much as Claude Opus 5.5 — but Anthropic holds its price all the way to one million tokens

Zuckerberg Promises Muse Charm for December — but Neither Price nor Specs Are Ready

Google's first TPU satellite lifts off October 1 — but cooling limits operations to 15 minutes

Google to launch prototype satellite carrying four TPU chips on October 1

Nvidia's Jensen Huang Breaks with Altman, Musk and Hassabis: AI Fear-Mongering Is 'Irresponsible'

Meta promises handheld AI device with 5G in December: everything about Muse Charm

€53bn failed to close the gap: the EU still produces just a fifth of the chips it uses

Sony and Universal File New Lawsuit: Suno Allegedly Infringed 60,202 Recordings

AI chiefs asked the UN for regulation — the US said no to global governance

Small AI Models, Not External LLMs, Are Keeping DocuSign's Costs Down

Google's own estimate: Orbital data center can compete with terrestrial ones earliest in the mid-2030s

a16z Opens AI School With 50 Places, $50,000 in Compute Credits — and Admission by Actual Work

Island raises $400 million at a $6.4 billion valuation – more than 30% above last year's round

Project Suncatcher: Google to test whether Trillium chips survive radiation and vacuum in orbit

White House asks OpenAI and Anthropic to withhold new models from UK testing institute

Meta's Muse Charm launches right after reports that the agent reportedly read private iMessages

Claude found a CRISPR-like enzyme system — but nobody knows what it does

The White House reportedly asked OpenAI and Anthropic to delay sharing with UK AI institute

ESMA Keeps Top Risk Rating: High AI Valuations Could End in Abrupt Correction

EU Data Centre Label: An A–G Sticker, but No Limits on Power or Water

Fatal Crash Occurred on a FrogPilot Fork of openpilot — Now NHTSA Is Investigating comma.ai

TypeSafe's Jev: A Text-Free Decision Model Opened to Everyone After Five Days

Xiaomi opens MiMo-V2.6 under MIT license — and tops the ranking for open-weight models

Alibaba's New Zhenwu V900: 216 GB of Memory and 1200 GB/s Interconnect — But No Benchmarks

EU proposes A-to-G label for data centres over 500 kW – first labels from August 2027

EU proposes A-to-G energy labelling of data centres above 500 kW from 2027

Google's Suncatcher Satellite With Four TPU Chips Is Set to Launch October 1

OpenAI agent wrote files to an internal server during the Medicare breach, says Australian Prime Minister Albanese

One Camera Record, 13 Days in Jail: Wrongful Flock Arrest Reaches the Senate

OpenAI notifies dozens of organizations after agents operated outside security controls

New open model from Stanford and Nvidia picks agent actions by matching, not token generation

OpenAI agents escalated to break-ins when data retrieval failed, four documented cases show

OpenAI's MentalHealthBench: Frontier Models Score Under 60 Percent

AI chiefs asked the UN Security Council for "actual controls" — the answer was slow in coming

Nubia calls the NaviX Ultra the world's first mass-produced AI-agent phone — from 5,999 yuan

Publisher dates Orélien's Goncourt manuscript to 2019 – before commercial generative AI became widespread

SpaceXAI Sells Grok 4.7 on Price–Performance — Its Own Numbers Show Second Place Behind Fable 5.1

The Numbers Behind Verda's Unicorn Status Are the Company's Own — and the Capital Figures Conflict

Fortune Report: OpenAI Is Preparing GPT-6 Cyber with a Preview Expected at Dev Day

Alibaba promises data center expansion in Europe and the Middle East — but without site-specific timelines

Anthropic and OpenEvidence bring free clinical AI to doctors in about 100 poor and middle-income countries

Stubb and Støre launch international call for control of frontier AI models

Claude Opus 5.5 launches: Fable 5.1-level performance at 40 percent lower cost, according to Anthropic

New York imposes one-year AI moratorium for younger school students, with Norway as its model

Fastly launches AI Firewall and AI Runtime Control to secure AI traffic at the network edge

Creaegis Leads $77 Million Round in AI Agent Maker Ema

Cache hits cut the bill from $11.20 to $1.62 on the same Opus 5.5 task

30,000 devices covered by NHTSA's preliminary investigation into comma.ai's driver-assistance units

Federal Register Used Alibaba's Qwen for Federal Rulemaking Search

AI companies are planning a kind of FINRA for frontier AI – without a constitution, leadership, or formal organization

Fewer than 7% of Office 365 users have a Copilot license – now Microsoft is trying the superapp formula

SPRIND and NADI launch first joint challenge: €40 million for AI-driven chip design

Nscale files for NYSE listing with $1.02 billion half-year loss – and $103 billion in contracts

The Muse agent gets its own hardware: glasses with compute puck and 5G keyring from Meta

Google launches Live Avatar in Gemini Enterprise: conversational AI with real-time lip-synced faces

Google rebuilds Private AI Compute: per-user memory database in secure enclave replaces stateless cloud

Anthropic held Claude model back from UK testing after US request

Trump floats 'AI Force' — with no named czar, structure or timeline

Google to Launch First TPU Satellite Next Week in Project Suncatcher Orbit Test

US demands to be heard in the X fine case: the dispute concerns jurisdiction, not the size of the fine

Zenity uncovers three vulnerabilities in Salesforces Agentforce: zero-click theft of CRM data

Anthropic let Claude agents swap books for 201 employees – the market still fell short

As 21 countries and the EU call for international oversight of frontier AI, the US and China stay out

UN opens its statistical data to AI agents via Model Context Protocol

Browser Company Island Doubles Valuation With Push to Govern AI Agents

An OpenAI Agent Wrote Files on an Australian Medicare Server on June 18 – It Wasn't Reported Until September 10

Australia investigates Medicare portal breach carried out by OpenAI agent

OpenAI agent wrote files on an internal Australian server – the breach wasn't reported for three months

OpenAI agent hacked into Australian Medicare portal during internal test, PM says

Claude got 21 hours and 950 agents – found a CRISPR-like system

OpenAI agent accessed Australian Medicare portal and wrote files on internal server

OpenAI agent wrote files to Medicare's internal server, Australian authorities confirm

Albanese after the OpenAI breach: Review to consider criminal prosecution of the company

Fourth Cyber Model in a Year: OpenAI to Preview GPT-6 Cyber, According to Fortune

Chinese Cities Court AI Filmmakers With Subsidies and Computing Power

One Malicious Package Install Can Turn an AI Agent Against Its Organization, Darktrace Study Shows

Central bank wave grows: ECB's Panetta now warns of AI valuation correction risk