OpenAI's Hallucination Fix Still Fails 3% of the Time
OpenAI just shipped receipts that its models are getting less delusional, but the real story is what those numbers reveal about how far we still have to go.
Continue reading ›Anthropic
posts
THE WIRE:
All
AI Agent EconomyAgents
Real World AssetsRWA
Human ImperativeHumans
The SignalSignal
Wire WeeklyWeekly
Meta faces publisher lawsuit that could redefine AI training costs
The training data question just got a price tag, and Meta's betting it won't stick. Five major publishers and author Scott Turow sued Meta for allegedly scraping books and journals from pirate sites like LibGen and Sci-Hub to train Llama models without permission
Continue reading ›
OpenAI Just Made Its Smartest Model Free for Everyone
OpenAI just made its smartest model yet the one everyone gets for free. GPT-5.5 Instant replaces GPT-3.5 Instant as ChatGPT's default model, giving hundreds of millions of free-tier users access to significantly more capable AI.
Continue reading ›
CopilotKit Scores $27M to End the Era of Dumb AI Chatbots
The fastest way to lose the AI agent war is to make developers choose between building their own stack from scratch or bolting on a chatbot that doesn't know what your app does.
Continue reading ›
Goldman Sachs Ditches Big Tech Cloud for In-House AI Agents
Wall Street just made its biggest bet that AI belongs in-house, not in the cloud—and they're betting against Big Tech owning the distribution.
Continue reading ›
US Government Gets Pre-Release Access to Google and Microsoft AI Models
The frontier model release schedule just got a new gatekeeper, and it's not a safety board or a nonprofit watchdog.
Continue reading ›
Wall Street's New Survival Strategy: Can't Beat Big Tech, So Buy In
When Big Tech comes for your margin, Wall Street doesn't compete—it joins forces, then brings the entire portfolio along for the ride.
Continue reading ›
US Government Gets First Look at Every Major AI Model Before Public Release
The companies building AGI just handed Washington the keys before anyone else gets to turn the ignition. Google, Microsoft, and xAI agreed to give the US government pre-release access to their AI models for security evaluation before public deployment.
Continue reading ›
OpenAI Commits $1 Trillion While Admitting Its AI Could Fail
The frontier AI race just turned into a game of corporate chicken played with nine-figure data center bills.
Continue reading ›
Google's AI Researchers Just Unionized Against Their Own Company's Military Deals
The researchers building the most powerful AI models in the world just decided their bosses can't be trusted with them.
Continue reading ›
Google's AI Scientists Just Unionized Over Pentagon Work
The people building the world's most advanced AI just decided they need collective bargaining rights to control what it does. Google DeepMind workers in the UK voted to unionize, requesting joint representation by the Communication Workers Union and Unite the Union
Continue reading ›
White House Wants Veto Power Over Anthropic's Next AI Release
When the Pentagon and the White House can't agree on whether an AI model is safe to deploy, you're watching the invention of a new regulatory regime in real time.
Continue reading ›
Google Made $46B on Anthropic While You Watched ChatGPT
The world's biggest AI arms dealer just made $46 billion while everyone was watching ChatGPT. Google owns 14% of Anthropic — worth roughly $53 billion on Anthropic's latest $380 billion valuation — after investing just over $4 billion total
Continue reading ›
Pentagon Tells Anthropic to Keep Its Safety Theater—Picks Rivals Who Actually Deploy
The Department of Defense just told the AI safety movement exactly what it thinks of red lines: find someone else to draw them.
Continue reading ›
OpenMythos Reverse-Engineers Anthropic's Most Dangerous Banned AI Model
The safest way to ensure a dangerous AI model stays locked away might be to build it yourself. OpenMythos is an open-source project attempting to reverse-engineer the architecture of Claude Mythos, Anthropic's cyber-capable model the company deemed too dangerous to release.
Continue reading ›
Cerebras IPO Valued Higher Than Most AI Chipmakers Despite Zero Name Recognition
The chip company OpenAI can't stop talking about just priced its IPO at a valuation that dwarfs most of the AI infrastructure stack — and barely anyone outside the Valley knows what they make. Cerebras, an AI chip maker with deep OpenAI ties, is targeting a $26.6B+ valuation in its upcoming IPO
Continue reading ›
Amazon's Own Engineers Rejected Its AI Tools for Claude and Codex
The world's most dangerous tech employer just admitted its own AI tools weren't cutting it. Amazon is rolling out Anthropic's Claude Code immediately and OpenAI's Codex on May 12 to all corporate employees, moving beyond its in-house Kiro tool after internal pushback
Continue reading ›
Anthropic Called Fully AI Employees in 2025. They Were Right.
The AI hype machine just got a year-older receipt, and it's more instructive than the original prediction.
Continue reading ›
Salesforce's Own Architect Just Raised $950M to Replace Salesforce
The enterprise AI layer just got its first credible challenger to Salesforce, and it's backed by the people who built Salesforce. Sierra raised $950M at a $4.5B valuation to become the infrastructure layer for AI-powered customer service, bringing total capital raised past $1B
Continue reading ›
GitHub's AI Explosion: 14× More Code Nobody Would've Written Before
The explosion isn't in how much code humans write — it's in how much code now exists that nobody would have bothered to write at all. GitHub commits are projected to hit 14 billion in 2026, up from 1 billion in 2025 — a 14× increase year over year as AI code generation becomes default infrastructure
Continue reading ›