Anthropic Launches Claude Chrome Extension to 1,000 Max-Plan Subscribers with Safety Tweaks

TechCrunch •

Anthropic rolled out a research preview of Claude for Chrome to 1,000 Max-plan subscribers ($100–$200/mo) and opened a waitlist. The Chrome extension runs a sidecar agent that keeps browser context and — with user permission — can take actions for you. Anthropic has default site blocks and prompts for high-risk tasks, and says safety tweaks cut prompt-injection success from 23.6% to 11.2% amid a fierce browser AI race.

Read original ↗

Also mentioned in:

  • Ars Technica — Anthropic’s Claude for Chrome Vulnerable to Prompt-Injection Attacks, Experts Warn