AI news for Wednesday, 18 November 2026
The day’s AI news in one list, the biggest stories first. Each story links to its short read and to the outlets that reported it.
- 1 · Funding & DealsAnthropic prices IPO at $83, above projected range
Anthropic, an AI company that created the Claude assistant, priced its initial public offering at $83 per share, above its projected range of $70 to $78.
Reported by TechCrunch and 1 more
- Microsoft makes Copilot the center of work with Office apps inside
Microsoft announced at its Ignite event that Word, Excel, and PowerPoint are now integrated inside Copilot, which the company describes as "the new OS for work." This reverses the traditional relationship where Office applications were…
Reported by Ars Technica and 1 more
- 3 · Policy & DramaNvidia is abandoning China after Jensen Huang’s ouster
Nvidia's new acting CEO announced the company will stop selling advanced chips in China, ending a previous arrangement that had allowed Chinese access to these products through a tariff-based loophole.
Reported by TechCrunch and 1 more
- Introducing Gemini 4 Neon
Gemini 4 Neon is a new AI model designed to handle complex tasks across software engineering, knowledge work, and scientific research.
Reported by Google DeepMind
- Rumble-2-Flash speech model launches for 40+ languages
Amsterdam-based Reson8 released Rumble-2-Flash, a 500 million-parameter open-source speech recognition model available for free download.
Reported by Hugging Face and 1 more
- Satya Nadella predicts AI agents will reshuffle the economics of the web
Microsoft's CEO predicts that AI agents will fundamentally change how the internet economy works, particularly affecting companies whose business models depend on advertising revenue and human web traffic.
Reported by TechCrunch
- 7 · Policy & DramaTrump administration restricts China's AI chip access
The Trump administration announced new restrictions making it harder for Chinese companies to access advanced AI chips by renting computing power from overseas data centers, a practice that previously bypassed existing export controls.
Reported by Wired AI
- 8 · Policy & DramaPalmer Luckey warns AI-powered cyberattacks are outpacing defenses: ‘It’s time to start running’
AI-powered cyberattacks are advancing faster than the outdated technology systems that defend against them, according to a warning from a tech executive at a recent conference. U.S.
Reported by TechCrunch
- 9 · Funding & DealsAnthropic increases compute spending 4x per year
Anthropic is spending money on computing resources at a rate of 4 times per year, while OpenAI's spending is growing at roughly 3 times per year.
Reported by Epoch AI
- 10 · Hardware & ComputeNvidia AI Surge Comes with $118 Billion in Guarantees
Nvidia reported record data center revenue growth while taking on up to $118 billion in financial guarantees to help customers build the power and facilities needed for AI infrastructure deployment.
Reported by Data Center Knowledge
- 11 · Policy & DramaOpenAI explains frontier AI risk management approach
Frontier AI models are rapidly becoming more capable, and organizations developing them need systems to manage safety risks across training, testing, and deployment stages.
Reported by OpenAI
- 12 · Agents & ProductsSnapchat lets developers build AR Lenses with AI coding assistants
Snapchat introduced Easy Lens CLI, a tool that lets developers build AR Lenses using AI coding assistants they already use, such as Claude Code and Cursor, rather than learning Snapchat's own development tools.
Reported by TechCrunch
- 13 · Models & ReleasesGenie 4 is now available
Google DeepMind has released Genie 4, an updated world model that generates interactive environments users can explore and edit in real time through their browser.
Reported by Google DeepMind
- 14 · BenchmarksAgentic inference capacity grew 4.6 times in a year
Between Q2 2025 and Q2 2026, the agentic inference capacity of AI chips sold grew 4.6 times, compared with 2.6 times growth in overall computing power.
Reported by Epoch AI
- 15 · Hardware & ComputeAgentic AI shifts demand toward central processors
Agentic AI is shifting demand away from graphics processing units toward central processing units as a key component of AI infrastructure.
Reported by SiliconANGLE
Every story, by topic
One short email a day with the stories that matter. Free. Unsubscribe with one click.