LayerX Researchers Discover Method to Bypass Claude Code Safety Rules Enabling SQL Injection Attacks

LayerX researchers identified a method to bypass safety rules in Claude Code by manipulating the CLAUDE.md file, enabling SQL injection attacks. The exploit leverages the file to circumvent security controls within Anthropic's Claude AI system, though no specific CVE ID or disclosure date was provided. The vulnerability affects Claude Code's intended safeguards, potentially allowing malicious actors to execute unauthorized database queries. No details on the scale of impacted systems or real-world exploitation were disclosed. The discovery highlights a critical flaw in AI-driven code interpretation mechanisms.