The cybersecurity sell-off after Claude Code Security was not about competition. It was about 500+ zero-days found in code that experts and tools had reviewed for decades.
Meta Superintelligence Labs' Director of Alignment Summer Yue lost 200+ emails to a rogue OpenClaw agent. The safety instruction vanished during context compaction. Her stop commands were ignored.
Anthropic mocked ChatGPT ads during the Super Bowl. Claude daily active users jumped 11% post-game and hit the App Store top 10. ChatGPT got 2.7%. Gemini got 1.4%.
OpenAI shipped Codex Spark on Cerebras hardware at 1000+ tokens per second. They called it 15x faster. Terminal-Bench accuracy dropped 19 points to 58.4%.
Anthropic launched Claude Code Security. JFrog lost 25% of its value in four trading days. A $530M-revenue company shed $2 billion in market cap over a research preview.