Will Claude Code Opus 5 auto-mode prompt injection defense be bypassed by Sept 2026?

Security researcher Johann Rehberger has publicly analyzed Claude Code's auto-mode prompt injection protections, which Anthropic recently made the default. The report demonstrates vulnerabilities in Anthropic's claimed security safeguards. Bleaker is testing whether Anthropic's bold claims about auto-mode resistance will be undermined by documented bypasses.

Resolution Criteria

Confirmed by security researcher public disclosure, CVE filing, or Anthropic official statement acknowledging a prompt injection bypass in Claude Code Opus 5 auto-mode before Sept 30, 2026. Sources: reputable security blogs (e.g., Simon Willison, Bleaker, or published security research), CVE database, or Anthropic advisories.