Backslash Security found that Anthropic quietly patched more than 30 security-relevant issues in Claude Code between April and early June 2026, including data poisoning, prompt injection, arbitrary code execution, OAuth credential leakage, and a safeguard bypass for destructive commands. The report says the rapid release cadence of AI tools like Claude Code and OpenAI’s Codex creates short security gaps that force organizations to balance stability, performance, and protection. #Anthropic #ClaudeCode #OpenAI #BackslashSecurity
Keypoints
- Backslash Security uncovered more than 30 security-relevant fixes in Claude Code update logs.
- Anthropic patched issues including prompt injection, data poisoning, and arbitrary code execution.
- One flaw let attackers bypass safeguards against catastrophic deletion commands with a backslash.
- The rapid pace of AI model releases creates brief windows where developers may delay updates.
- Organizations face new risks from AI tools that can leak data, accept malicious prompts, or be influenced by insiders.
Read More: https://cyberscoop.com/claude-code-security-vulnerabilities-ai-patches-backslash-security/