Anthropic AI Updates: March 30, 2026

1. Claude Paid Subscriptions More Than Double

Anthropic confirmed that Claude’s paid subscriptions have more than doubled in 2026, according to a TechCrunch analysis of billions of anonymized credit card transactions from approximately 28 million US consumers. The growth is attributed to heightened public attention from the Pentagon dispute, Super Bowl advertising, the Claude Code developer tool, and Computer Use capabilities, with the majority of new subscribers joining the $20/month Pro tier. Source

2. Claude Autonomously Discovers Ghost CMS Zero-Day in 90 Minutes

Anthropic researcher Nicolas Carlini published results showing Claude autonomously discovered a blind SQL injection vulnerability in Ghost, a publishing platform with 50,000 GitHub stars and a historically clean security record. The vulnerability allowed an unauthenticated attacker to compromise the admin database, create an admin account, and take full control of any Ghost installation — and Claude found it in 90 minutes without human guidance. The demonstration underscores that AI systems are already capable of identifying novel vulnerabilities at a speed that challenges traditional security research timelines. Source