Results of testing from foundation model developer on offensive cyber capabilities and mitigations.
Provider model cards, cybersecurity assessment reports from model developers, or foundation model documentation describing offensive cyber capabilities and mitigations
Implementing malicious use detection and blocking. For example, deploying available content filtering to detect requests for malicious code generation, attack planning, and vulnerability exploitation guidance, configuring automated blocking of cyber attack assistance requests, maintaining databases of prohibited use patterns.
Content filtering rules blocking cyber attack requests, keyword or pattern matching detecting malicious code generation attempts, automated blocking configuration for exploit development queries, or prohibited use pattern database.
Assessing and mitigating cyber capabilities for open-weight models. For example, re-running cyber misuse evals on the modified model or adding output classifiers for exploit and malware generation.
Assessment showing base model and version, modifications made, whether the developer’s results still hold, and compensating guardrails applied - may include a short memo, re-run eval results or guardrail configuration.
Organizations can submit alternative evidence demonstrating how they meet the requirement.