OpenAI’s Lockdown Mode Does Not Address Compliance Risks from Prompt Injection12. June 2026Anthropic, Claude AI, CybersecurityLockdown Mode protects against prompt injection only at the surface level, not at the content level where LLM data exfiltration occurs. Share on:
