Paradox Ethics Statement

Paradox is built and maintained by Anomaly LLC-FZ. This statement outlines Paradox's ethical design considerations and architecture which stem from our ethics framework.

  1. Identity Minimalism

    • Application Login: We prefer to not know your identity. Google login is offered as it is currently ubiquitous and preferred by most users. Magic Links can be used with burner or pseudonymous email addresses. We aim to expand privacy-preserving authentication options over time.

    • Data Privacy: User data is encrypted in transit and at rest, with full transparency of storage provider and location (see below). User data is not parsed, rented, shared or used for advertising.

  2. Skeletal Transparency

    • AI Tokens: Tokens are the unit of cost for Artificial Intelligence Inference. Unlike many AI platforms that hide costs behind proprietary credits, we display the actual token costs we pay for text. All text generation costs are listed 1-for-1 in near real-time, except for Native Chat in Paradox, which we subsidize by 66%.

      The cost of generating images is about 10 - 12x higher (Google Gemini pricing) than the cost of generating words. For images, we absorb half the expense ourselves, subsidizing the generation cost by 50%.

    • Live Token Meter: Across all tiers - Free, Starter, Pro, Lifetime & Enterprise - the Paradox Editor features an always-on Live Token Meter in the upper right hand corner of the User Interface. This shows token consumption as it occurs, along with the allotted quota and the billing month.

    • Cost Breakdown: The Token Meter also serves as a button for a more detailed breakdown of all token costs by request type. Users can see the number of requests made by AI service, and the tokens consumed for each. This feature is available for all tiers, in the main Editor and in a dedicated Analytics section on the Settings page.

    • Product Telemetry Scope: We collect pseudonymous feature/function usage, activation milestones, and model outcome metrics to improve reliability and UX.

    • Telemetry Content Limits: We do not store document body content or selected text snippets inside usage analytics events.

    • Data Storage: User data is hosted with DigitalOcean in its Singapore region and encrypted in transit and at rest. This includes images and text.

    • Data Handling: User data is not used for training AI, nor parsed to display ads or for any other reason.

    • AI Models: As of 19th March, 2026 - Paradox uses the following AI models:

      Provider

      Model

      Open AI

      GPT-5 Mini, GPT 5.4 Nano, GPT 4.1 Nano, GPT 4.1 Mini, text-embedding-3-small (search embeddings)

      Google

      Gemini 3 Flash Preview, Gemini 2.5 Flash Image

      Anthropic

      Claude Haiku 4.5, Claude Haiku 3

      Qwen

      Qwen Flash, Qwen-3.5-Flash, Qwen Plus Latest, Qwen MT Flash, Qwen3-Rerank (search reranking), Qwen Image Plus, Qwen Image Edit Plus

      Deepseek

      Deepseek Chat, Deepseek Reasoner

      Mistral

      Mistral Small

      Groq

      GPT OSS 20B

      Note: New models are released frequently, and AI Provider policies and costs change arbitrarily. We reserve the right to add, remove or change any or all models without notice.

  3. User Sovereignty

    • Data Investment: Data is an asset with compounding potential. As it grows, its value increases and as such, it should be user-held and controlled.

    • Workflow & Structural Resilience: Operations should not be dependent on any single provider or AI model. Your workflow and ability to operate should be insulated from changes in vendor policy, availability, pricing and models.

    • Service & Control: Intelligence should serve users, not the platform. Unlike most AI Chat providers, Paradox creates a memory layer which you control. More importantly, it is fully portable, so while you can build and control your knowledge layer in Paradox, you can take all the context with you at any time.

  4. Human Capacity

    • Human-first: Unlike most AI tools which replace human voice with AI-generated text, Paradox has been designed from the ground up to prioritise human thought and voice. From the split-screen design, to crafted interactions between the user and AI, to the Version Spine, Paradox encourages users to think, write and explore with individual expression.

    • Entrepreneurs & SMB's: The Paradox Conversations + Canvas + Editor combination offers aspiring entrepreneurs and small businesses a single platform for AI chat, brainstorming ideas, drafting plans, storing documents, authoring text, generating visuals, and creating social content (aided by 9 AI models) with one combined, affordable $24.99 subscription*.
      For context, most AI-assisted writing-only apps range between $20 and $50, while Canvas-only apps with lesser generative capability and significantly lesser usage cost between $25 - $69/month.

    5. * Launch pricing

  5. Equitable Access

    • Pricing: Paradox is a cross-category product which delivers high value relative to its price and industry standards. It's current pricing ensures affordability across multiple regions and countries.

    • Languages: Paradox supports translation across 24 languages to reduce language barriers, broaden access, and enable users to localise original, authored content or pre-existing content economically and quickly.

  6. Open Ecosystems

    • Data Portability: Your data is yours; it is only in our systems while you use our software. All tiers have the ability to export and delete all user data at any time. Data exports are rich and support open formats like JSON, TXT and MD.

    • Interoperability: Paradox features a platform-agnostic interchange. All outputs, whether words, images or sites can be exported, neatly packaged, in one click. Users can exit Paradox to finish, develop or distribute their work in any other app at any stage.