ChatSSC and other SecurityScorecard AI features may occasionally return inaccurate or incomplete information. This is a known limitation of generative AI systems, commonly referred to as an AI hallucination.
Hallucinations can occur when a model lacks sufficient or up-to-date information, misinterprets context, or cannot confidently infer an answer. As a result, you should treat AI-generated responses as assistive and always verify outputs before taking action.
To reduce the likelihood of hallucinations, SecurityScorecard employs several safeguards:
Continuous Monitoring
We use AI evaluations to flag responses outside the expected range. We also monitor and log key information, while respecting data privacy, to quickly identify and address potential issues.AI Guardrails
Protective measures are in place that prevent misuse and ensure responses adhere to SecurityScorecard guidelines.User Feedback Loop
Users can provide feedback directly in features like ChatSSC by thumbs-downing responses that are inaccurate or unhelpful. You can also reach out to Support with any questions or feedback.
Learn more about AI and data practices at SecurityScorecard, see the AI Model and Data Overview, or our AI Legal Addendum.