Skip to main content
OpenEye Knowledge Base

OpenEye AI Visual Search and AI Visual Check - Privacy, Security, and Responsibility Use Overview FAQ

OpenEye uses AI in video security and operations workflows to help users work more efficiently, surfacing relevant video faster and enabling automated monitoring of business, safety, and operational conditions. These features are built on enterprise-grade AI infrastructure and are designed with privacy, security, and responsible use as foundational requirements, not afterthoughts.

Feature Overview

What is AI Visual Search?

AI Visual Search helps users locate relevant recorded video using natural language descriptions. Users submit a text query and the system surfaces matching video results.

What is AI Visual Check?

AI Visual Check analyzes camera snapshots to answer business, safety, and operational questions. It evaluates a single image against a configured prompt and returns a match, no-match, or indeterminate result.

Does AI Visual Search identify specific people?

No. AI Visual Search processes text-only queries. No customer images are sent to the large language model (LLM). It does not perform identity recognition.

Does AI Visual Search use facial recognition?

No. AI Visual Search does not use facial recognition. It is a text-based search feature. Biometric Face Re-ID is a separate, distinct feature with its own strict regional controls.

Data and Processing

What data is processed by AI Visual Search?

AI Visual Search processes the user’s search text when a search is run. Images associated with events are processed separately in OpenEye’s cloud when those events occur; they are not sent to a third-party AI provider as part of the search. OpenEye also records search activity.

What data is processed by AI Visual Check?

AI Visual Check processes a camera snapshot, a natural-language prompt (canned or custom), and an optional reference image or text. OpenEye also records prompts, usage information, status, and timing for tracking.

Does OpenEye use a third-party AI provider?

Yes. OpenEye uses third-party AI providers and also runs models within its cloud environment. OpenEye uses enterprise API tiers with contractual data protections for its third-party AI providers.

If a third-party provider is used, what data is sent?

For AI Visual Check, the image and prompt are sent to the provider. For AI Visual Search, text-only queries are sent and no customer images are included. Data shared with providers is limited to what is necessary to fulfill the request.

Is customer data used to train AI models?

No. Our AI providers contractually guarantee that customer inputs and outputs are never used to train, retrain, or fine-tune their foundation models. Prompts, responses, and related data are never used for training and are never shared with model creators or other customers.

Retention and Deletion

How long is AI-related data retained?

Retention periods vary by data type:

  1. AI usage records are retained for 12 months.
  2. AI Visual Check activity records and AI Visual Search chat logs are retained indefinitely.
  3. Images processed during checks are subject to a scheduled cleanup, typically within 10 days.
  4. Detection and search-related data in the cloud is retained between 10 and 60 days depending on the data type.
  5. Third-party AI providers do not retain customer data after processing is complete.
Access and Security

Can employees access customer data?

Neither OpenEye nor its AI providers perform human review of customer data. Abuse detection is fully automated. Internally, sensitive data is never logged, and biometric PII is stripped from all system logs. Production data access for AI feature data follows the same existing policy as all other production access.

What permissions are required?

All AI features are restricted to authenticated users with the appropriate user group permissions. Every AI request requires an authenticated user. AI Visual Check requires the appropriate alert management permissions to save or run a check. AI Visual Search requires the appropriate search and camera access permissions, with results filtered to the user’s authorized scope.

Are actions audited?

Yes. OpenEye logs the full AI Visual Check job lifecycle and prompt-moderation results. A record is persisted for each AI interaction, capturing the account, user, AI provider, usage details, use case, timestamp, and requester. Provider-side logging is also available for audit purposes.

Guardrails and Limitations

What guardrails exist?

AI Visual Check moderates prompts before they are saved and rejects flagged content, including abuse, off-topic requests, and jailbreak attempts. Both features support AI self-reported moderation flags. Rate limiting is enforced through hourly and daily usage budgets and per-request caps. The risk of someone trying to manipulate the AI is minimal by design because the AI model cannot take direct action within the system, and results are always filtered to the user’s authorized scope.

What are the known limitations?

Results are AI-assisted, not guaranteed. Each AI Visual Check evaluates a single snapshot with no video or temporal analysis. Indeterminate results occur when an image is insufficient for a determination. Accuracy depends on the configured AI model.

Customer Responsibilities and Documentation

What does the customer remain responsible for?

Customers are responsible for acknowledging region-specific disclosure and consent language in the UI before activating AI features. For biometric features such as Face Re-ID, customers must ensure compliance with applicable local laws. Customers are also responsible for their own use of AI outputs and for ensuring appropriate human review of AI-assisted decisions.

What documentation is available under NDA?

Additional documentation that may be available under NDA includes security questionnaires, Data Processing Agreements (DPA), SOC reports, and penetration test summaries. Contact your OpenEye representative for details.

Is customer feedback used for product improvement?

Yes, but only when explicitly submitted. When a user submits feedback on an AI Visual Search result or AI Visual Check alert through the OWS feedback mechanism, the associated materials may be used by OpenEye to improve its features and test coverage. This is opt-in only and is distinct from the third-party model training guarantees described above. Note that explicitly submitted feedback content may be transferred to the United States.

  • Was this article helpful?