Upskillable

Responsible AI & data use

Last updated: November 15, 2026 · Version 2

This overview summarizes our November 2026 AI Policy and customer agreement. Those contractual documents govern customer processing; this page does not change an existing agreement.

Where AI is used

The AI Policy identifies nightly administrator insights, natural-language reporting and the learner-portal assistant as core AI functions. Other AI and media modules depend on the customer's selected features. Relevant prompts, source material and context may be sent to the providers involved in those functions. AI processing is not synonymous with model training.

Improving the service

The agreement permits anonymized, aggregated usage statistics to operate and improve the service where neither the customer nor an individual can reasonably be identified. Tenant-specific indexing and embeddings used to deliver the service are also permitted.

Anonymization must prevent identification; removing names, hashing identifiers or pseudonymizing records is not sufficient on its own. Identified usage and billing records remain protected data.

Model training

The current customer agreement prohibits Upskillable and its subprocessors from using Customer Data, prompts or AI Outputs to train or improve shared or general-purpose models. An exception requires a signed amendment. This overview does not create an exception for a future anonymization or training process.

A proposal to improve models using anonymized information must first establish the permitted purpose, effective anonymization, provider terms and applicable contractual approvals. We do not represent a planned anonymization or training process as a verified live capability.

No-training commitments do not mean zero retention. Provider retention depends on the provider, service and account configuration; request the applicable details during your security review.

Understanding learner memory

The learner portal includes a Memory Trust Center where learners can review information used by the tutor and confirm, edit or forget individual memory facts. These controls relate to the learner's AI memory within their organization; they are not a request to erase every account, learning or security record.

For broader access, correction, export or erasure requests, contact your organization's administrator or security@upskillable.ai. The Privacy Policy explains our roles and the request process. Account deactivation should not be understood as complete data erasure.

Human review and accountability

AI output may be inaccurate or biased. Customers review it before publishing or relying on it and apply meaningful human review to significant decisions. Voice and likeness features require the person's documented consent.

Agent Actions are subject to the purchased scope and approval requirements in the Order Form and AI Policy. Uploaded documents or model output do not grant an agent authority.

Questions and documents

Contact security@upskillable.ai for the AI Policy, applicable provider information and processing questions. The Subprocessor List distinguishes core services from optional feature providers.

A few practical questions

Responsible AI & data use, in plain language.

Are customer prompts, documents, or AI outputs used to train shared AI models?

No. Upskillable’s customer agreement prohibits Upskillable and its subprocessors from using customer data, prompts, or AI outputs to train or improve shared or general-purpose models without a signed amendment.

What is the learner Memory Trust Center?

The Memory Trust Center inside the learner portal lets individual learners inspect preferences or context remembered by the AI Coach and confirm, edit, or forget memory facts at any time.

Which AI functions run automatically versus on administrator review?

Administrators and authors review and approve generated lessons, updates, and Agent Action proposals; customer agreements require human review before significant certification or employment decisions.

Does Upskillable require consent before voice or likeness cloning?

Yes. Documented consent from the person whose voice or likeness is used is required under Upskillable’s AI Policy before activating voice or presenter cloning features.

Back to Trust Centre