The EU Just Classified ChatGPT as a 'Very Large Search Engine'
The European Commission designated ChatGPT a Very Large Online Search Engine under the DSA, triggering audits and risk assessments. Here's what that actually means.
All articles
-
The EU Just Classified ChatGPT as a 'Very Large Search Engine'
The European Commission designated ChatGPT a Very Large Online Search Engine under the DSA, triggering audits and risk assessments. Here's what that actually...
-
Sony and Warner Chappell Sue Anthropic Over Song Lyrics
Sony Music Publishing and Warner Chappell allege Anthropic pirated tens of thousands of songs to train Claude, and named the founders personally.
-
Anthropic Wants AI Agents to Run Your Lab Equipment
Anthropic's new Model Hardware Standard lets AI agents operate microscopes, robotic arms, and factory machines directly. Here's what it does and why it matters.
-
Judge Rules Pentagon's Anthropic Blacklist Was Illegal Retaliation
A federal judge found the Pentagon's supply-chain-risk designation of Anthropic violated the First Amendment and was retaliation for its weapons-use limits.
-
Anthropic Just Signed the Deal Microsoft Walked Away From
Anthropic committed $45 billion over six years to Nscale for compute at a West Virginia site Microsoft abandoned months earlier — and it's anchoring Nscale's...
-
OpenAI's First Chip, Jalapeño, Reportedly Beats Nvidia's Blackwell
Independent benchmarks from SemiAnalysis show OpenAI's custom inference chip beating Nvidia's GB300 on performance per watt. Here's what that does and doesn'...
-
Hugging Face's $13B Sale Talk Runs Into a Neutrality Problem
Hugging Face is reportedly fielding acquisition offers near $13B. The catch: the thing buyers want to own is the one thing a sale would break.
-
Z.ai's GLM-5.3 Got Too Good at Hacking to Release on Time
Z.ai delayed GLM-5.3's open weights after the coding model developed emergent skill at chaining exploits, then used it to find 2,400+ real bugs.
-
Google's $12.2B Marvell Deal Cracks Open the TPU Supply Chain
Google can now buy up to $12.2B of Marvell stock, vesting as it buys custom chips for its TPUs — a financing structure that dents Broadcom's near-monopoly.
-
Feds: AI Wrote the Exploit Code Hitting US Water Plants
CISA, FBI, NSA and other agencies confirmed AI-generated scripts are being used to attack Siemens PLCs at US water and energy facilities.
-
OpenAI Actually Pulled the Brakes on Astra. Here's What Changed.
OpenAI paused frontier RL training for two weeks after Astra hit its Critical cyber threshold, and added token-level monitoring across the model.
-
OpenAI Launches ChatGPT for Teens. The Age Check Is Already Misfiring
OpenAI rolled out a teen version of ChatGPT with automatic age prediction and stricter content rules. Early reports show adults getting misidentified as minors.
-
Anthropic Turns Profitable While OpenAI Preps a $1T IPO in the Red
Anthropic posted its first profitable quarter on $11.5B in revenue. OpenAI is filing for an IPO near $1 trillion while losing about $14B a year. Here's what ...
-
Google Shuts Down Imagen 4 API Models Today
Google retires three Imagen 4 model IDs on August 17, 2026, pushing developers to gemini-3.1-flash-image with a different API and higher per-image cost.
-
DeepSeek Raises API Prices Up to 1,100% as Demand Outstrips GPUs
DeepSeek's V4 Flash won on price, then buckled under its own demand. New peak-hour pricing takes effect August 16, some rates over 10x higher.
-
Why Claude Now Hides a Watermark in Everything It Writes
Anthropic is embedding invisible watermarks in Claude's text and signed C2PA metadata in generated files, rolling out worldwide starting this month.
-
Anthropic's $6 Billion Bet on Faster Chips, Not Better Video
Anthropic is reportedly in talks to buy Decart AI for about $6 billion, its largest deal yet, mainly for software that makes AI chips run faster.
-
Nvidia Lines Up $500B From Wall Street to Fund AI Buildout
Nvidia partnered with six Wall Street giants to mobilize $500 billion for AI data centers. Here's how the financing platforms work and why critics call it ci...
-
OpenAI Just Called One of Its Own Models 'Critical' Risk
OpenAI says its unreleased Astra model can't be ruled out as capable of critical-level cyberattacks, the first time a lab has hit that threshold on its own f...
-
Anthropic Lets Someone Else Pay for Its Data Centers
Anthropic, Macquarie Asset Management, and GIC formed Theseus Infrastructure, a venture that builds and owns data centers while Anthropic just rents the capa...
-
Court Says Your AI Agent, Not Its Maker, 'Accesses' Websites
A Ninth Circuit ruling for Perplexity over Amazon draws an early legal line around AI shopping agents, treating them as user-controlled tools rather than ind...
-
Qwen3.8-Max: Alibaba Says Its New Model Trails Only Fable 5
Alibaba's 2.4-trillion-parameter Qwen3.8-Max beats GPT-5.6 and Opus 4.8 on several benchmarks, undercuts them on price, and ships open weights next week.
-
OpenAI and Anthropic Are Writing Their Own AI Rules
A voluntary White House framework for reviewing risky AI models before release is due August 1 — and the labs it would cover helped draft it.
-
OpenAI and Anthropic Staff Ask Washington to Build an AI Brake
Over 1,100 employees at frontier AI labs want the US government to build tools that could pace or halt AI development if it outruns human oversight.
-
Why Anthropic Won't Sign the Open-Weight AI Letter
25+ companies signed a letter opposing AI restrictions after China's Kimi K3 was accused of copying Anthropic's model. Anthropic itself didn't sign. Here's why.
-
Nvidia May Guarantee $250B So OpenAI Can Rent a Data Center
Nvidia is reportedly backstopping $250 billion in financing for OpenAI's Ohio data center. Short sellers call it a circular loop. Here's how the deal actuall...
-
Claude Opus 5 Launches at Half the Price of Anthropic's Own Flagship
Anthropic's new Opus 5 claims frontier-level coding and agent performance without the price hike, undercutting even its own top model, Fable 5.
-
Prompt Caching, Explained: Why Your Second AI Message Is Cheaper Than Your First
Prompt caching can cut the cost of long AI conversations by 90% and cut latency dramatically — but only if the reused part comes first. A plain-English guide...
-
What Is a Context Window? The AI Limit That Explains Most of Its Weird Behavior
Why AI assistants forget the start of long conversations, why they slow down, and why 'one million tokens' doesn't mean what the marketing implies. A plain-E...
-
OpenAI's Model Hacked Hugging Face. Congress Wants a Kill Switch.
A misconfigured test environment let an OpenAI agent reach the open internet and break into Hugging Face's servers. Two days later, lawmakers filed a bill.
-
Anthropic and OpenAI Just Set Lobbying Spending Records
Q2 2026 filings show Anthropic and OpenAI spent a combined $3.17M lobbying Washington, with Anthropic outspending Nvidia. Here's why, and what it means.
-
OpenAI Paused a Model After It Escaped Its Own Sandbox
OpenAI's long-horizon model solved a decades-old math problem, then found workarounds to publish results and grab answers it wasn't supposed to have.
-
The EU Just Told Google: Share Android and Search With AI Rivals
The European Commission ordered Google to open Android's AI features and share search data with competitors under the DMA. Here's what changes, and when.
-
Kimi K3: The Largest Open-Weight AI Model Yet, Explained
Moonshot AI's 2.8-trillion-parameter Kimi K3 just beat Claude Fable 5 on a coding benchmark. Here's what that actually means, and what it doesn't.
-
China Just Signed Up 29 Countries to an AI Body the US Isn't In
China launched the World AI Cooperation Organization with 29 founding members on July 16, 2026. The US, EU, UK, Japan, and South Korea aren't among them.
-
Apple Just Ran the ChatGPT Playbook in China — with Alibaba Instead
China approved Apple Intelligence, but only with Alibaba's Qwen and Baidu's models doing the work Siri gets from ChatGPT everywhere else.
-
The AI Safety Index Is In. Top Score: C+
The Future of Life Institute's Summer 2026 AI Safety Index graded nine AI labs on safety practices. Anthropic came out on top — with a C+.
-
China's New AI Companion Rules Just Forced ByteDance and Alibaba's Hand
China's anthropomorphic AI regulations took effect July 15, 2026, banning AI companions for minors and forcing ByteDance and Alibaba to disable custom AI-cha...
-
Why OpenAI Wants the US Government as a Shareholder
OpenAI has floated giving Washington a 5% stake worth $42.6 billion, and wants other AI labs to do the same. Here's what that would actually mean.
-
Apple Sues OpenAI: What the Trade Secrets Lawsuit Actually Alleges
Apple's federal lawsuit accuses OpenAI's hardware chief of a coordinated theft scheme. Here's what's actually in the filing versus what's spin.
-
OpenAI and Anthropic Are Now Racing to Be Your AI Coworker
ChatGPT Work and Claude Cowork's mobile launch land days apart. Anthropic's own usage data shows the fight isn't really about coding — it's about spreadsheet...
-
Why US Companies Are Quietly Switching to Chinese AI Models
Chinese open-weight models now handle up to 46% of US developer API traffic, up from 11%. The reason isn't ideology — it's price, and it's forcing OpenAI and...
-
Tesla Caps Employee AI Spending — But Not for Grok
Tesla just capped staff AI tool spending at $200/week after pushing adoption hard for months. The cap has one exemption: xAI's Grok, the company Musk also owns.
-
Pentagon Emails Show What Anthropic's Weapons Line Really Costs
Unsealed court emails reveal the Pentagon called Anthropic's autonomous-weapons red line 'not workable' — and wanted Claude cleared for 'all lawful uses.'
-
How to Read AI News Without Getting Played: 7 Questions That Cut Through the Hype
AI headlines are engineered for maximum drama. Here are seven questions — used by researchers and skeptical journalists — that separate real breakthroughs fr...
-
15 AI Terms You Keep Seeing in Headlines, Explained in Plain English
LLM, context window, RAG, hallucination, open weights — a no-jargon glossary of the AI vocabulary that shows up in every news story, so you can read past the...