Anthropic launched Claude Opus 5 at the same price as its predecessor, but with results nearly double on real-world work benchmarks. Here's what changed and why it matters for enterprise AI costs.
AMD unveiled Helios, its first full AI server rack packing 72 Instinct MI455X chips. Anthropic has ordered 2 gigawatts of capacity, and Microsoft is building Azure around it.
The White House accused Moonshot of stealing Anthropic's tech to build Kimi K3, and Treasury threatened sanctions and export blacklisting. The Chinese firm is also accused of dodging a ban on Nvidia chips via Thailand.
Governor Pritzker signed a law forcing OpenAI, Anthropic, Google, Meta and xAI to hire independent auditors every year to check their models' safety. Violators face fines of up to $3 million.
The Claude-maker signed a 20-year lease with TeraWulf for a 401 MW campus in Kentucky worth $19 billion. A former Bitcoin miner just became the infrastructure backbone for one of the world's top AI companies.
A researcher found obfuscated code in Claude Code that silently detected Chinese users via timezone checks and steganographic prompt changes. Anthropic removed it July 1, one day after public exposure.
Anthropic's most capable public model is running again. Claude Fable 5 returned on July 1 — two and a half weeks after it was pulled worldwide over sudden US export controls tied to a security incident.
Anthropic released Claude Sonnet 5 on June 30, a model that nearly matches Opus 4.8 on agentic benchmarks while costing developers less than half as much — starting at $2 per million tokens.
Two weeks after banning both models, the US reversed course for Mythos 5: 100+ companies and agencies can now access it. Fable 5 stays fully blocked. The episode created the first government 'trusted tier' for AI.
From April to June 2026, Alibaba used 25,000 fake accounts to run 28.8 million exchanges against Claude — harvesting its reasoning to train Qwen. Anthropic calls it the largest distillation attack ever filed.
OpenRouter launched Fusion — a compound AI that fires prompts at Gemini, DeepSeek, and Kimi in parallel. On benchmarks, it nearly matches the banned Fable 5 at half the price.
John Jumper, Nobel laureate and AlphaFold creator, is leaving DeepMind for Anthropic. Days earlier, Transformer co-author and Gemini co-lead Noam Shazeer left Google for OpenAI.
On June 12, the US ordered Anthropic to disable Fable 5 and Mythos 5 globally under export controls. The alleged jailbreak behind the order is a security technique Anthropic says GPT-5.5 can already do.
Anthropic released Claude Fable 5 — its most capable widely available model, with Mythos-level power but built-in safeguards. Alongside it comes Claude Mythos 5, unrestricted but only for vetted customers.
An unreleased version of Claude, codenamed Oceanus, reached red-team testers and within hours was being resold through a Chinese proxy. Anthropic has reportedly paused its testing program while it investigates.
Microsoft researchers caught a prompt injection flaw in Claude Code's GitHub Action — attackers could trick the AI into leaking API keys and CI/CD secrets from any workflow. Patched on May 5.
A bipartisan bill released June 4 proposes the first comprehensive U.S. AI framework: mandatory audits for frontier labs and a three-year ban on states passing new AI development laws.
Six Anthropic engineers are deployed inside the NSA using Mythos for offensive cyber operations — on the same day the company publicly called for a global halt to advanced AI development.
Anthropic expanded Project Glasswing to 150 organizations across 15+ countries — including NATO, ENISA, Samsung, SK Telecom, and Okta — to use Mythos AI for hunting vulnerabilities in critical infrastructure.
Anthropic, the maker of the Claude AI assistant, has filed to go public and could list as early as this fall. Its valuation nears $965 billion and run-rate revenue has topped $47 billion.
