Home AI Article

Anthropic Blocks Third-Party Claude Access, Cuts Competitor API Keys

TL;DR

Anthropic revoked unauthorized harness access to Claude models and blocked xAI from using Claude via Cursor, forcing high-volume automation toward metered API pricing.

Key Points

  • Anthropic tightened safeguards against spoofing Claude Code client identity, severing third-party tools like OpenCode from consumer subscription pricing
  • xAI lost access to Claude models via Cursor IDE after using them to train competing systems, violating commercial terms
  • Pattern established: OpenAI revoked in August 2025, Windsurf blocked in June 2025, now xAI and third-party wrappers targeted simultaneously
  • Users exploited flat-rate Claude Pro ($200/month) with autonomous agents consuming tokens worth $1000+ monthly on metered API pricing

Why It Matters

Developers relying on third-party integrations to access Claude at consumer rates must immediately migrate to official APIs with per-token billing or face sudden access loss. This reflects a broader consolidation strategy where AI labs enforce strict API boundaries, making 'shadow AI' infrastructure a compliance and operational risk for enterprises.
Full breakdown on VentureBeat

Source: venturebeat.com