Legitimate Interests Assessment

Date: 2026-11-06. Controller: Vig Solutions LLC. Privacy contact: info@trainingos.com (the published Training OS privacy contact).

Purpose

Suggest skills taught or assessed by the creator’s own eligible public course content and maintain the Allure skills graph. This helps creators describe their content and users find relevant skills. It does not assess learners, establish mastery, advertise to individuals, or train general-purpose models.

Necessity

Activity titles alone can omit what a course teaches. A bounded outline of the creator’s authored lesson text, headings, captions, image descriptions and quiz questions supplies the relevant context without sending the original archive to the model. Only supported captured original uploads are eligible; shared playback packages are not recovered under another creator’s identity. Access is scoped to the verified creator and destination tenant, with bounded extraction and processing limits.

Balancing and safeguards

Creators retain ordinary hosting and playback without analysis. Existing creators receive an email and dashboard notice and a full 30-day period before their default can apply; an explicit early opt-in is available. New creators must accept the updated Terms/Privacy at signup and see the current product notice. Existing opt-outs remain in force; creator and per-upload opt-outs, tenant restrictions and operator holds are enforced before capture, processing and result storage. No learner activity, learner assessment results or learner exports are used. Sensitive personal information and private, passphrase-protected, NDA, archived or unpublished content are excluded. Authored text can still name third parties: creators must have authority to process it and omit sensitive or confidential material; a creator’s upload cannot authorize processing on someone else’s behalf.

Only bounded outlines and proposed mappings are retained for review, not a certification or verified mastery. Turning analysis off stops new analysis, queues withdrawal of earlier mappings and deletes private analysis-source copies through durable cleanup. Copies are also deleted no later than 12 months after an item is deleted or unpublished, whichever event comes first. Bounded outlines and mappings are kept while the content exists, subject to withdrawal and applicable rights. Audit and security records follow the existing schedule. External processing already completed cannot be undone. OpenAI’s published API data controls, including default training exclusion and retention exceptions, are linked in the privacy notice; zero retention is not assumed.

Outcome and review

Legitimate interests is the basis for this narrow purpose, not consent. Notice acknowledgment, default settings and an early preference choice are not GDPR consent. On this scope and these safeguards, the creator/content interests are balanced against the limited necessary processing. Hold processing where authority, eligibility or notice evidence is missing. Reassess before expanding inputs, purposes, recipients or access, or if complaints show the balance has changed. Rights requests: info@trainingos.com.