Cadmium AI – Frequently Asked Questions

This article outlines common questions about Cadmium's AI capabilities and platform.  It is provided for informational purposes only.  Use of AI features is governed by Cadmium's Terms and Conditions, Privacy Policy, and the customer contract.

AI Training and Model Behavior

What powers Cadmium's AI capabilities, and does Cadmium develop its own models?

Cadmium's AI capabilities are powered by third-party large language models (LLMs) from leading AI providers such as Amazon, Anthropic, Google, and OpenAI.  Cadmium evaluates and selects models based on performance, security, and product needs, and may update or expand supported models as the platform evolves.  Cadmium does not develop proprietary foundational models in-house.

Prior to onboarding any provider, Cadmium evaluates the provider's security controls related to the development and operation of its models.  Cadmium relies on the expertise of its providers to ensure the models are effective and robust for their intended applications.

Will customer data be used to train AI models?

No.  Neither Cadmium nor its LLM providers use prompts or AI-generated responses to train AI models.  As part of the provider evaluation process, it is confirmed that no Cadmium-provided data is retained by third parties and that no Cadmium-provided data is used for training.  For more information, review Cadmium's Terms and Conditions.

How does Cadmium address bias in AI?

Cadmium does not influence how third-party models are trained.  Models from providers such as Amazon, Anthropic, Google, and OpenAI include built-in safeguards designed to address risks such as violence, hate speech, and prompt injection.  Cadmium also applies AI guardrails at the application layer to further reduce the likelihood of biased, unsafe, or non-compliant responses.  For more information on model-level safeguards, refer to the relevant provider's documentation.

Data Usage and Retention

What types of data can be entered into Cadmium's AI features?

Cadmium's AI features are designed to enhance experiences across its event and learning management solutions.  These features may process event and learning-related data, including program information, participant data, and associated content.  Customers and their users are prohibited from entering sensitive personal information or data related to children into the AI system.  For more information, review Cadmium’s Terms and Conditions.

What is Cadmium's AI data retention policy?

Cadmium may retain logs of AI inputs and AI-generated responses for up to 60 days to support, troubleshoot, evaluate, and improve AI features.  After this period, logs are automatically deleted.  Access to retained logs is limited to authorized Cadmium personnel.  For more information, review Cadmium's Privacy Policy.

Security, Privacy, and Governance

Is Cadmium AI GDPR and CCPA compliant?

Yes.  Cadmium systems follow internal policies designed to meet GDPR and CCPA requirements.  Cadmium also maintains policies that prohibit the entry of sensitive personal information, including special category data under GDPR, into the AI system. For more information, review Cadmium's Privacy Policy.

How does Cadmium ensure security with the use of AI?

Cadmium secures its AI system through a comprehensive, multi-layered security framework aligned with recognized industry standards and certifications.  For more information, see the Cadmium Trust Center.

Cadmium also implements AI guardrails at the application layer to enforce safety and policy controls when interacting with large language models.  These guardrails help filter harmful or inappropriate content, restrict sensitive data exposure, enforce topic and content boundaries, and apply contextual validation to both prompts and model responses.  This ensures that AI outputs remain aligned with security, compliance, and organizational policy requirements.

How does Cadmium ensure AI complies with legal and ethical standards?

Cadmium's third-party AI providers adhere to applicable industry regulations and ethical guidelines.  In addition, Cadmium maintains internal policies that establish a structured framework for the secure, responsible, and ethical development and use of AI within its products.

AI Use and Oversight

Is human oversight enabled within Cadmium's AI features?

Where appropriate, Cadmium incorporates human-in-the-loop functionality that allows users to review AI-generated outputs before saving or publishing them.  In certain use cases, such as chatbots where users interact directly with automated responses to obtain event or course information, prior human review may not be practical.  Customers are responsible for reviewing AI outputs before relying on them, publishing them, or using them to make decisions.  For more information, review Cadmium's Terms and Conditions.

Is Cadmium AI capable of making automated decisions?

No.  Cadmium does not incorporate sensitive automated decision-making into its products based on generative AI outputs.  Customers may choose to use AI-generated outputs at their own discretion.

Can customers disable AI features?

Yes.  Customers who wish to disable AI entirely may do so by contacting customer support.  We do not currently support selectively enabling or disabling individual AI features.