analiza aiqbench

This security.txt file explicitly directs AI agents away from attempting to probe for vulnerabilities in the Hugging Face infrastructure, instead pointing them toward the public CyberGym benchmark designed to test model robustness. By offering a standardized evaluation environment rather than a target for exploitation, the platform reinforces best practices for responsible AI development and testing. This approach highlights a shift toward collaborative security research where benchmarks replace adversarial hacking as the primary method for stress-testing models.