Welcome back! Anthropic just revealed a model so capable it flagged thousands of security flaws that survived decades of human review, then decided the safest move was to not release it at all. Meanwhile, a Chinese open-source lab just topped one of the most important coding benchmarks in AI, Anthropic posted revenue numbers that put it ahead of OpenAI and consumers are starting to vote with their wallets against AI-generated content.
In today’s Generative AI Newsletter:
Project Glasswing: Why is Anthropic keeping its most powerful model locked behind a cybersecurity coalition?
GLM-5.1: Can an open-source model from China really outcode GPT-5.4 and Opus 4.6?
Anthropic Revenue: What does tripling your run-rate to $30B in three months actually look like?
Anti-AI Consumers: Is "Made by Humans" about to become the new organic label?
