Legal
AI Training Disclosure
This disclosure explains how active beta uploads and derived AI data may be used internally to evaluate and improve StuntIQ models, and what deletion changes.
StuntIQ AI Training Disclosure
Version: 1.1 Effective Date: March 2026 Applies to: All StuntIQ users during the closed beta
Purpose of This Disclosure
StuntIQ uses uploaded beta videos and derived AI data to evaluate and improve its models. This disclosure explains what that means in the current beta and what deletion changes.
1. What the AI Pipeline Does
When a video is uploaded, StuntIQ's automated pipeline may:
- detect people in the video
- track movement over time
- extract pose/keypoint data
- classify athletic roles and stunt phases
- segment attempts
- estimate difficulty and generate analysis artifacts for the product UI
These outputs are experimental and may change as the beta evolves.
2. What Data May Be Used for Model Improvement
While a video remains active in the beta, StuntIQ may use the following internally for model evaluation and improvement:
- pose/keypoint data
- attempt features and attempt metrics
- phase, role, and other AI labels
- AI run metadata and confidence values
- related review and correction signals generated inside the product
Raw video footage is not shared with external model-training providers.
3. How This Data Is Handled
- Model-improvement use is internal to SkillGrade Labs LLC.
- We do not sell user video data or derived AI data for training purposes.
- We do not promise indefinite retention of anonymized or aggregated training features in this beta disclosure.
- We do not provide a separate product control that keeps a video active for analysis while excluding it from future internal model-evaluation use.
If you no longer want a specific uploaded video to remain in active model-evaluation use, request deletion or consent revocation for that video.
4. What Deletion Changes
When a specific uploaded video is deleted, or consent is revoked for that specific video, StuntIQ removes the application-managed active data for that video from the product, including the stored video, derived AI artifacts, run records, pose/keypoint data, and per-run dataset export files tied to that video.
Deletion does not retroactively retrain, edit, or undo models that were already trained before the deletion completed.
5. Beta Context
The closed beta is intended to improve StuntIQ's model quality using real-world sports video data. Participation therefore includes internal use of active uploaded videos and their derived data for model evaluation and product improvement.
6. Questions
For questions about training-data use or specific-video deletion:
SkillGrade Labs LLC Email: privacy@skillgradelabs.com