Security built around explicit context.
AskAny.ai is designed so AI work starts with a user action: select text, attach a file, crop a screenshot, speak a prompt, or type a request. The trust model is simple: make context intentional, routing visible, and BYOK controls available when subscribers prefer their own provider account.
What teams can verify first
AskAny.ai processes context the user actively submits.
AskAny.ai does not train its own foundation models on customer prompts, files, screenshots, or chats.
Paid subscribers can route eligible model traffic through supported personal provider keys.
Subprocessors, model-provider notes, privacy details, and data-flow documentation.
User-triggered
AI requests start only when someone submits selected content or a prompt.
Visible routes
Users can see and understand which model and provider route is being used.
BYOK controls
Paid subscribers can add supported provider keys and choose when eligible requests use personal routing.
Reviewable assets
Trust pages explain how selected context, providers, and account settings are handled.
From selected context to answer, step by step.
The security boundary is easiest to understand when you can see where data enters, where it is prepared, and which route processes it.
User chooses context
Text, files, cropped screenshot, voice, or typed prompt.
AskAny.ai prepares request
Context, model choice, task instructions, and required metadata.
Route is checked
AskAny.Ai checks the selected model and active routing configuration before sending the request.
Model inference
The selected provider or customer-key route processes the request.
Answer returns
Users can copy, replace, save, continue, or discard the output.
A practical security posture for desktop AI.
AskAny.ai focuses on practical controls for individual use: intentional context capture, visible routing, provider behavior, and personal BYOK.
Review model providers and routingContext controls
- •User-triggered capture, not continuous recording.
- •Region-based screenshot capture is visible to the user.
- •Users submit only the text, files, voice, or prompts they choose.
Provider controls
- •Official model APIs where supported.
- •No-training options configured where available.
- •Provider matrix documents handling differences.
Personal BYOK controls
- •Supported provider keys and explicit model IDs.
- •Requests use the selected personal route when BYOK is active.
- •Provider-side billing, logging, and retention settings continue to apply.
Transparency resources
- •Subprocessor list and model-provider details.
- •Privacy policy and request data-flow documentation.
- •Clear descriptions of hosted routes and personal BYOK.
Security FAQ: the hard questions
Do you have SOC 2 or ISO 27001 certification?+
Not yet. AskAny.ai currently provides a security overview, subprocessor list, provider documentation, and architecture/data-flow information. Formal certification may be considered as the product grows.
Can I route supported traffic through my own keys?+
BYOK is available to paid subscribers for supported providers. Eligible model traffic can use your own provider account so that provider’s billing, logging, and retention settings apply.
What happens when a user submits a confidential document?+
The selected document content is processed to generate the requested answer and may be sent to the selected model provider. Users should submit only data they are authorized to process. Users can choose supported BYOK routes when they want provider-account controls to apply.
Do third-party AI providers retain prompts?+
Provider handling varies by provider, plan, and contract. AskAny.ai maintains a provider matrix and configures no-training options where supported. BYOK customers should review and configure their own provider accounts.
Can AskAny.ai be self-hosted?+
Self-hosting is not currently offered.