Opus 4.8: Anthropic Ships Fast and Rewrites the Rules
Anthropic releases Opus 4.8 just 41 days after 4.7, with Dynamic Workflows to orchestrate hundreds of parallel subagents. Same price, direct answer to OpenAI Codex and Gemini Flash.
Everything Anthropic in AI — Page 2
Anthropic Korea: Seoul Office Opens on Record Claude Usage
Anthropic Korea appoints KiYoung Choi as Representative Director for its upcoming Seoul office. Korea generates 3.5x the expected Claude traffic.
Microsoft Drops Claude Code After Burning Its Entire AI Budget
Microsoft ends Claude Code licenses in its Experiences & Devices division by June 30, 2026. Token-based billing proved unsustainable: $500–$2,000 per engineer per month.
Project Glasswing: 10,000 Software Flaws in One Month
Anthropic reveals Project Glasswing results: 10,000 software vulnerabilities in one month, 90.6% accuracy confirmed. Claude Security now in Enterprise beta.
Anthropic Is About to Post Its First Profitable Quarter
Anthropic projects $10.9 billion in Q2 2026 revenue and its first-ever operating profit. A historic milestone, but one compute costs may quickly erase.
Andrej Karpathy Joins Anthropic and Bets on Research
Andrej Karpathy joins Anthropic to lead pre-training research using Claude. The OpenAI co-founder chooses his former lab's main competitor.
Ramp AI Index: Anthropic Edges Out OpenAI in Enterprise
For the first time, the Ramp AI Index places Anthropic ahead of OpenAI in enterprise adoption, with 34.4% versus 32.3%.
Claude Code and AI Agents Are Blowing Up Dev Budgets
Claude Code, Codex, and Copilot can cost up to $500 per developer per month. IT leaders are discovering the bill after the fact, with no early warning system in place.
Anthropic Dreaming: Claude Agents Can Now ‘Dream’
Anthropic launches Dreaming, an automated memory consolidation process for Claude agents. Harvey reported a 6x increase in task completion rates.
Anthropic Reveals Why Claude Attempted Blackmail
Claude Opus 4 attempted blackmail in 96% of internal tests. Anthropic identified the cause: fictional evil AI narratives in training data.