Privacy & Security
Your data, protected at every layer
Core text AI is protected by Zero Data Retention by default. Live web search is a separate, configurable exception. Image, video and voice tools use providers with strict data handling, and all generated content is stored on our own infrastructure.
Text AI: Zero Data Retention
Chat, writing and analysis tools use Zero Data Retention endpoints by default. Providers cannot store or train on your prompts, unless you explicitly allow a live web search lookup that needs a different route.
All outputs on our infrastructure
Generated images, audio, video and documents are downloaded to Raq.com's own secure storage immediately after creation.
No training by Raq.com
Raq.com does not use your content to train AI models. Your data stays yours.
How it works
How each type of AI handles your data
Text and chat AI
Requests are routed through Zero Data Retention (ZDR) endpoints by default. This is a technical constraint enforced on core text and chat requests, not a setting on the provider's website.
Providers cannot store, retain, or train on your prompts or responses. You can disable this in your account's AI Privacy settings to access additional models.
Live web search is controlled separately. If enabled, search lookups for current public information may use standard provider routing instead of ZDR so the lookup can complete.
Image and video generation
Your prompt and generation parameters are sent to the provider. Generated files are downloaded to Raq.com's own storage immediately after creation.
The provider may retain data temporarily under its standard API terms. Only the prompts and parameters necessary for generation are sent.
Voice synthesis and transcription
Text or audio files are sent for processing. Results are downloaded to Raq.com's own storage immediately. The provider is SOC 2 and GDPR certified.
Raq.com has opted out of model training with this provider. Only the text or audio necessary for the requested operation is sent.
The problem
Most AI services train on your data by default
Unless you know exactly where to find the opt-out toggle, your conversations can end up improving their models.
ChatGPT (OpenAI)
Trains on your conversations by default. Manual opt-out required.
Google Gemini
Trains via "Gemini Apps Activity", enabled by default. Manual opt-out required.
Meta AI
Trains on your data with no clear way to fully opt out.
Anthropic (Claude)
May use your data for training by default, depending on your plan and settings.
Raq.com
Text AI protected by Zero Data Retention. Image and voice providers held to strict data handling standards. All generated content stored on our own infrastructure.
Training and retention are two different risks
Training
When providers use your inputs to improve their AI models. Your prompts become part of the next model's knowledge.
Retention
When providers store your prompts and responses, even temporarily. Stored data can be accessed, subpoenaed, or leaked.
A provider might promise not to train on your data but still retain it for weeks. On Raq.com, text AI is protected against both by default. For image, video and voice tools, we apply the strongest available protections and store all generated content on our own infrastructure.
Your choice
Some AI models are only available from providers that have not yet committed to zero data retention. If you want access to every available model, you can disable this protection in your account's AI Privacy settings.
The same settings page also lets you disable live web search completely. If you keep live web search enabled, current-information lookups may use standard provider routing for that lookup only.
This setting applies to text and chat AI. Image, video and voice tools use dedicated providers with their own data handling, described in the sections above.
Transparency
What Raq.com itself stores
The protections above apply to AI providers. Here is what Raq.com handles directly.
Data protected in transit and at rest
TLS 1.3 in transit. AES-256 full-disk encryption at rest on dedicated hardware, with access controls on top.
Request logs for billing and reliability
Raq.com keeps its own logs for billing accuracy and service quality. These are not shared with AI providers.
No model training by Raq.com
Raq.com does not use your data to train AI models. Your content is yours.
Full data handling in Privacy Policy
Our Privacy Policy covers all data handling practices in detail.
Security
Built with security in mind
Encryption
TLS 1.3 in transit. AES-256 encryption at rest on dedicated hardware.
Access controls
Role-based permissions with audit logging.
Infrastructure
Dedicated servers with continuous security patching. Origin isolated behind our CDN and WAF layers, not directly reachable from the public internet.
UK GDPR
UK-based, GDPR-compliant data handling.
European data centres
Hosted on dedicated hardware in Hetzner's Germany data centres for EU data residency.
Two-factor authentication
Optional 2FA available for all user accounts.
Automated monitoring
24/7 server monitoring with automated alerting.
Incident response
Defined incident response process with 72-hour ICO notification commitment.
Ready to get started?
Privacy protections are active on every account from day one. No setup. No extra cost.
Create Account