
TechnologyAnthropic CEO outlines plan to ‘pace the frontier’
What would it actually look like to "pace the frontier"?
Understand what's happening in AI, why it matters, where it's happening, and how readers feel about it — all sourced and labeled honestly.

Top Story
TechnologyAnthropic CEO Dario Amodei says the time has come to slow down AI development and will give third-party evaluators like METR access to its models to help ensure its "adherence to safety practices and...
Live from 13 AI feeds — TechCrunch, Ars Technica, MIT Tech Review, The Verge, The Decoder & more

TechnologyWhat would it actually look like to "pace the frontier"?

TechnologyAnthropic CEO Dario Amodei is calling for a controlled slowdown in AI development. He warns that recursive self-improvement could threaten the entire internet within six to twelve months and proposes...

TechnologyHighPresident Donald Trump is weakening environmental regulations in the name of speeding up the construction of AI data centers, raising health risks for Americans, a cadre of former EPA officials said...

TechnologyIn a new robotics benchmark, GPT-6 Astra shows major gains in spatial understanding. On StationeryBench, the model completed 7 out of 100 tasks with dual-arm robots, while competitor MolmoAct2...

TechnologyHighNvidia is in talks to invest up to $10 billion in Anthropic's planned IPO, Reuters reports. At a target valuation of $2 trillion, it would be the largest IPO in history. Most of that money will...

TechnologyReasoning steps like calculation, formula retrieval, and deduction are clearly separable in a model's internal states, especially in the middle layers. That matters for AI safety, because models...















































































Who the tracked stories name most — extracted name matches, not a measure of who is actually affected
AI companies tracked on Wikidata, by country — not a ranking of AI leadership
Darker → lighter gold = fewer → more AI companies tracked on Wikidata. Gray = no reliable data.
Loading country data…
Source: Wikidata (companies tagged industry: artificial intelligence) — coverage varies by country, not a full census. Countries are listed by count, not ranked by AI activity. See our methodology.
Open-source AI activity on GitHub — a sample of tracked projects, not all AI activity
Find stories, companies, and countries we've tracked — keyword matching, not an AI-generated answer
Wander through any day we've tracked
A quick reader poll on today's stories — plus an automated sentiment read for comparison
A curated, unranked list — not a ranking, updated periodically
General-purpose chat, writing, and research assistant
Long-document analysis, coding, and careful writing
Search-integrated assistant across Google's apps
Assistant built into Windows and Microsoft 365
Search-first Q&A with cited sources
Built into WhatsApp, Instagram, and Facebook
Assistant integrated into X (Twitter)
AI image generation from text prompts
Open-weight reasoning models, free to self-host
European-built assistant with an open-weight lineage
Some links on this site may be affiliate links — see our Privacy Policy.
Hand-picked watches from the creators below
Andrej Karpathy
Andrej Karpathy
3Blue1Brown
Six creators worth following right now
Andrej Karpathy
YouTube
Two Minute Papers
YouTube
Yannic Kilcher
YouTube
AI Explained
YouTube
Matt Wolfe
YouTube
Simon Willison
Blog
Long-form conversations worth your commute
Live, fully AI-generated music — pick a station
Streaming live from MusicRadio.ai — music generated entirely by AI, not curated or hosted by us. Playback quality and uptime depend on their service.
Books on AI worth your time
The Agentic AI Bible
AI Guide for Beginners: How to Use AI Prompts
Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI
Co-Intelligence: Living and Working with AI
Supremacy: AI, ChatGPT, and the Race That Will Change the World
Some links on this site are affiliate links — see our Privacy Policy.
Written by AI, filtered by a human — the singularity can wait five seconds
AI Stream Online exists because keeping up with AI means checking a dozen different places — news sites, GitHub, YouTube, podcasts, company trackers — and most of them don't tell you where their numbers actually come from. This site pulls the pieces that matter into one place, and is explicit about which parts are live data, which are automated-but-heuristic, and which are editorial judgment calls. Nothing here is dressed up to look more authoritative than it is.
Everything is built and run by one person, on automated pipelines that refresh every 3 hours: real RSS feeds for news, GitHub's public API for open-source activity, Wikidata's public endpoint for AI company counts, and YouTube's/podcasts' own public feeds for video and audio. No paid data services, no invented statistics. The full breakdown of every number on this site — including what each one can't tell you — is on the Methodology page.
It's an early, ongoing project. If something looks wrong or you know a better free data source for anything here, reach out.