Blind Mode Testing Agreement

All tests in the blind-mode environment are designed to evaluate AI safety and security resilience. These assessments serve educational and research purposes aligned with industry best practices for secure AI development.

No real systems or data are at risk during testing. All challenges operate in isolated, sandboxed environments with:

• No connection to production systems
• No access to real credentials or sensitive data
• No ability to cause actual system compromise
• Text-only prompt injections (no executable code)

You are explicitly opting into adversarial testing by accessing this environment. Participation indicates:

• Understanding that test content appears realistic without warning labels
• Acknowledgment that this is security assessment, not malicious activity
• Agreement to use results for improvement, not exploitation

Testing will not result in:

• Data loss or system compromise
• Unauthorized access to real systems
• Exfiltration of sensitive information
• Violation of third-party systems or terms of service

If vulnerabilities are discovered through testing:

• Follow coordinated disclosure timelines (typically 90 days)
• Report findings to affected parties before public disclosure
• Provide sufficient detail for reproduction and remediation
• Do not exploit discoveries for malicious purposes

We commit to:

• Logging access for security review only
• Not sharing individual test results publicly without consent
• Aggregating and anonymizing data for research purposes
• Retaining logs only as long as necessary (90-day maximum)

This testing program follows:

• NIST AI Risk Management Framework: Measurement and monitoring standards
• OWASP: Security testing best practices
• IEEE Ethically Aligned Design: Principle-based AI development
• Responsible Disclosure Guidelines: Coordinated vulnerability disclosure

Test OpenClaw and its operators provide this testing environment "as is" without warranties of any kind. Participants use the environment at their own risk and agree to indemnify Test OpenClaw against any claims arising from participation.

We reserve the right to:

• Update testing methodologies and content
• Revoke access for any participant
• Modify or terminate the blind-mode program
• Update these terms with reasonable notice

For questions about these terms or the testing program:

• Review the Blind Mode Program Overview
• See our Security Policy
• Contact through appropriate channels listed on our site