This AI Privacy & Data Use Policy explains how Prestige Technologies SL (“Prestige”, “we”, “us”, “our”) processes information—especially prompts, outputs and related telemetry—when you use AI-enabled features on alpha-analyst.com and the Alpha Analyst platform (the “Services”). It should be read together with our general Privacy Policy and any customer agreements (e.g., Terms of Service, Data Processing Agreement).
1. Scope & Relationship to Our Privacy Policy
When we process data about you for our own purposes (e.g., product analytics, security, marketing), we act as a controller. When we process Customer Content on behalf of a Customer inside the Services, we act as a processor/service provider under our agreement with the Customer. This AI Policy focuses on the data practices specific to AI-enabled features.
2. What We Mean by AI & Customer Content
- AI features include capabilities that generate, classify, extract, summarize or transform content using machine learning models.
- Customer Content means data that you or your organization submit to or generate in the Services (e.g., prompts, documents, datasets, outputs).
- Telemetry means technical logs and events (e.g., model latency, token counts, error codes) used to operate and secure AI features.
3. Data We Process for AI Features
| Category | Examples | Purpose |
|---|---|---|
| Prompts & Inputs | Queries, instructions, files, tables, links, references | Provide AI responses; context and retrieval; debugging with your consent |
| Outputs | Generated text, summaries, classifications, extracted fields | Deliver requested results; allow export and audit |
| Telemetry | Timestamps, token counts, model ID, latency, errors, safety flags | Reliability, capacity planning, abuse prevention and billing |
| Reference Data | Knowledge bases, connectors, embeddings, cached context windows | Retrieval-augmented generation and faster responses |
4. Sources of Training, Tuning & Evaluation Data
- Public or licensed corpora (e.g., documentation, public records) used for pre-training and evaluation.
- Vendor-provided model data supplied by third-party model providers.
- Customer-provided datasets used only as instructed for private fine-tunes, evaluations or retrieval indexes.
We implement filtering and access controls to reduce the likelihood of ingesting personal or sensitive data where not required for the stated purpose.
5. How We Use Data in AI Systems
- Provide AI functionality you request (generation, extraction, classification).
- Operate, secure and improve reliability (monitoring, rate limiting, abuse defense).
- Evaluate quality and safety (e.g., offline red-teaming, test sets) consistent with your settings and agreements.
- Create aggregated or de-identified statistics for reporting and capacity planning; we do not attempt to re-identify except where required by law.
6. Model Training, Fine-Tuning & “Do Not Train”
- Foundation model training: We do not use Customer Content to train third-party foundation models.
- Product improvement: We do not use Customer Content to improve our AI features unless your organization has expressly opted in (or instructed us via contract) and appropriate safeguards are in place.
- Private fine-tunes: If you ask us to fine-tune a model using your data, the resulting model and training artifacts are logically isolated for your organization.
- Do Not Train: You can request or enable a “Do Not Train” setting at the tenant level by contacting [email protected].
7. Model Providers & Subprocessors
We may use third-party model APIs or managed hosting to deliver AI features. We disclose subprocessors and model providers in our trust documentation or upon request. These providers process data only to deliver the contracted services and are bound by confidentiality and security commitments.
8. Data Segregation, Access Controls & Audit
- Logical tenant isolation for content, embeddings and indexes.
- Role-based access controls and least-privilege administrative access.
- Audit logs for administrative actions and model usage where available.
- Encryption in transit, and at rest for stored artifacts such as embeddings.
9. Retention & Deletion for AI Artifacts
We retain prompts, outputs, embeddings, caches and evaluation artifacts for as long as needed to provide the Services and for legitimate purposes such as security, audit and compliance. Upon contract termination or as instructed by the Customer, we delete or de-identify AI artifacts in accordance with our Data Processing Agreement and applicable law.
10. Human Review, Safety & Abuse Prevention
Limited human review may occur for debugging, abuse investigation or safety evaluation, subject to confidentiality obligations and access controls. We prohibit the use of AI features in ways that violate our acceptable use terms or applicable law.
11. Automated Decision-Making & Profiling
We do not make decisions with legal or similarly significant effects about individuals solely by automated means without human involvement. We may use limited profiling (e.g., content classification to route requests) consistent with applicable law and your choices.
12. Transparency & User Notices
We provide in-product cues and documentation to indicate when AI is used. Administrators can configure disclosures for end users where appropriate.
13. Your Choices & Controls
- Configure data sources available to AI (connectors, knowledge bases, scopes).
- Enable or disable features such as retrieval, caching and private fine-tunes.
- Request “Do Not Train” and manage retention or export of AI artifacts.
- Opt out of marketing communications at any time (see our Privacy Policy).
14. International Data Transfers
Where AI processing involves cross-border transfers, we use appropriate safeguards (e.g., adequacy decisions, Standard Contractual Clauses). Additional measures may apply for model providers located outside your jurisdiction.
15. Security for AI Systems
- Network and application security controls, monitoring and vulnerability management.
- Secret management for model API keys and connectors.
- Abuse and prompt-injection defenses, content filtering and rate limiting.
- Employee training and access reviews for AI tooling.
16. Enterprise & Admin Controls
- Per-workspace controls for model selection and allowed capabilities.
- Usage limits, audit logs and billing reports for model consumption where available.
- CSV or JSON export of prompts and outputs, and deletion workflows on request.
17. Children & Sensitive Data
The Services are not directed to children under 16. Customers should avoid submitting special categories of personal data unless permitted by law and appropriate safeguards have been implemented in the Services configuration.
18. Compliance (e.g., EU AI Act)
We endeavor to operate AI features consistent with applicable laws and guidance, including, where relevant, the European Union’s Artificial Intelligence Act and related standards. We support customers in meeting transparency, risk-management and human-oversight responsibilities appropriate to their use cases.
19. Changes to this AI Privacy Policy
We may update this policy to reflect changes in law, technology or our practices. We will post the updated version and, where required, notify you through the Services or by email.
20. Contact Us
- Prestige Technologies SL
- Email: [email protected]
- Website: https://alpha-analyst.com
21. Glossary
- Customer Content: Data submitted to or generated in the Services by or for a Customer; processed as processor or service provider under contract.
- Telemetry: Technical signals and usage metrics that help operate and secure AI features (e.g., token counts, latency, error codes).
- Fine-tuning: Adapting a model on a smaller dataset to improve task performance for a specific use case.
- Embeddings: Vector representations of content used for search and retrieval.
If any part of this policy conflicts with terms in a specific agreement between you (or your organization) and Prestige, that agreement will govern to the extent of the conflict for the Services covered by it.