Langflow CVE-2026-0768 exploited to steal OpenAI and AWS credentials

Affected Systems

Langflow versions 1.4.2 and earlier. The vulnerability exists in the code validator of the custom component editor's validate endpoint. Patched in version 1.11.6.

Exploitation Status

Active exploitation confirmed. VulnCheck observed 360+ attacks via honeypots, primarily from Russia. Attackers targeting LANGFLOW_SUPERUSER, OPENAI_API*, AWS_ACCESS*, AWS_SECRET* environment variables, /root/.cache/langflow/secret_key, SSH access, and bash history. No public PoC available.

Business Impact

Unauthenticated remote code execution with root privileges allows credential theft from AI application environments. Organizations using Langflow for AI workflows face immediate risk of OpenAI API key compromise, AWS credential exfiltration, and administrative account takeover. Langflow has been repeatedly exploited in 2026 (CVE-2026-33017, CVE-2026-5027, CVE-2026-55255, CVE-2026-0770, CVE-2026-9198), indicating sustained attacker interest.

Urgency

🔴 Immediate

Recommended Actions

• Upgrade all Langflow instances to version 1.11.6 immediately
• Rotate all OpenAI API keys, AWS access keys, and AWS secret keys used in Langflow environments
• Rotate Langflow superuser credentials and secret keys stored in /root/.cache/langflow/secret_key
• Review logs for validate endpoint access from unauthorized sources, especially from Russian IP ranges
• Audit environment variables and .ssh directories on Langflow servers for unauthorized access or modifications
• If immediate patching is not possible, restrict network access to Langflow instances to trusted IP ranges only