AI Disclosure

This AI Disclosure is issued by Neyda Technologies INC. (“we,” “our,” or “us”), the company that operates Jenora. It applies to all AI-powered features within the Jenora mobile application, web application, and related services (the “Services”).

1. Our AI Provider

Jenora uses the OpenAI API to power its AI features. All AI requests are routed through Jenora’s own secure backend server — the OpenAI API is never called directly from your device, and your API credentials are never exposed to the client. The specific model may change over time as we improve quality, safety, cost, or reliability. Material changes affecting how your data is processed will be disclosed before they take effect.

OpenAI does not use API-submitted data to train its models unless we explicitly opt in, which we do not. Depending on the endpoint and service configuration, OpenAI may retain API inputs and outputs temporarily for security and abuse monitoring purposes unless zero-data retention is approved and enabled for that endpoint.

AI requests are processed programmatically through Jenora’s secure backend server. To protect your privacy, these requests are stripped of account-identifying markers before being transmitted to OpenAI — your real name, email address, and user account identity are removed and replaced with generic placeholders. These controls are built directly into our backend request validation pipeline rather than relying on staff policy alone.

2. What AI Does in Jenora

AI is used for the following specific features in the Services:

Companion AI Chat

Jenora’s in-app companion responds to your messages with supportive, educational, and motivational responses. The AI draws on your personal context (mood, values, recovery status) to tailor its tone and content. It is designed to feel personal — but it is not a therapist and does not provide clinical advice.

Journal Reflection Summaries

When you complete a journal entry, AI can generate a brief reflective summary based on what you wrote — including mood themes, feeling patterns, and potential warning signs you identified. This summary is grounded entirely in your own words and is not embellished or extrapolated.

Behavioural Pattern Detection

Jenora’s behavioural pattern detection is primarily rule-based and statistical — patterns are computed directly from your check-in history, mood logs, and engagement data. AI may be used to help explain or summarise detected patterns in plain language, but it does not independently diagnose, classify disorders, or make clinical assessments. Patterns are surfaced as insights and are not stored as permanent diagnoses.

Structured Helpers

When you use tools such as SMART goal setting, cognitive reframing, or trigger identification exercises, AI assists in generating structured, step-by-step output in a validated format. These responses are constrained to the specific task and do not engage in open-ended conversation.

Tone-Matched Paraphrasing

Jenora can rephrase certain content in a tone that matches your preference (e.g., casual, empathetic, direct). This feature never paraphrases safety-critical content such as crisis messages or emergency hotline information.

Content Routing and Resolution

When you submit a message or query, the AI helps route your input to the most appropriate response — whether that is content from Jenora’s curated library (local, no AI generation involved) or a cloud-generated response. This routing happens automatically and silently.

3. What Data Is Sent to the AI

When a cloud AI feature is used, the following categories of your data may be included in the prompt sent to OpenAI, depending on the feature:

• Mood and wellbeing: Current mood rating and risk level indicator (low / moderate / elevated / high / critical)
• Recovery context: Days since sobriety start date and recovery path (if applicable)
• Values and goals: Your selected core values, value rankings, and active goals
• Health metrics (opt-in only): Sleep quality, step count, active minutes, and HRV score — only if you have enabled health data integration
• Journal content: Mood reflection, journal summary text, and feeling tags — for journal summarisation only
• Conversation history: Recent turns of an active AI conversation session to maintain coherent context
• Exercise input: Your text responses within a structured helper exercise

We do not send raw journal entries, therapy session notes, or therapist-linked data to OpenAI. Data sent to OpenAI is the minimum required to generate a relevant, personalised response.

Neyda Technologies personnel do not monitor or browse your AI conversations or journal summaries. A limited, authorised technical team member may review a specific AI input and output only if you explicitly initiate an in-app report or support request regarding a low-quality, inaccurate, or unsafe response. This access is strictly logged, audited, and bound by confidentiality obligations and internal access controls.

4. What AI Is Not Used For

• AI is not used to make automated decisions that have legal or similarly significant effects on you
• AI does not determine your access to the Services, pricing, or account status
• AI-generated content is not presented to therapists as clinical data — therapists see your raw check-ins, mood logs, and roadmap progress, not AI summaries
• AI is not used to infer diagnoses, classify disorders, or produce clinical assessments
• Your data is not used by OpenAI to train its foundation models under our API agreement
• AI is not used to target you with advertising

5. Safety Guardrails

We have implemented multiple layers of safety controls around all AI features:

• Local crisis detection: Before any message is sent to OpenAI, it is scanned locally on our servers for crisis keywords (suicidal ideation, self-harm, overdose signals). If detected, the AI is bypassed entirely and a crisis response with emergency contact information is returned immediately.
• Response filtering: AI-generated responses are validated before being shown to you. Responses that encourage harm, normalise substance misuse, or contain inappropriate content are blocked and replaced with a safe fallback.
• Crisis content protection: The paraphrasing feature is explicitly prohibited from rephrasing safety messages, crisis hotline numbers, or any content flagged as safety-critical.
• Prompt injection prevention: All user inputs are sanitised before being included in AI prompts. Instructions that attempt to override the AI’s behaviour (such as “ignore previous instructions”) are detected and blocked.
• Graceful degradation: If the AI service is unavailable or returns an unsafe response, the app falls back to Jenora’s curated local content library. You will always receive a response, even if cloud AI is offline.

6. Limitations of AI in Jenora

You should be aware of the following inherent limitations:

• Not clinically reviewed: AI-generated content is not reviewed by a licensed mental health professional before it is shown to you.
• May be inaccurate: AI can produce content that is factually incorrect, outdated, or contextually inappropriate for your specific circumstances.
• No full medical context: The AI does not know your full medical history, psychiatric history, or current treatment plan. Its responses reflect only the in-app data shared with it.
• Not a crisis service: The AI is not a substitute for human crisis intervention. If you are in distress, contact a crisis service directly.
• Potential for bias: Large language models can reflect biases present in their training data. We actively monitor and work to mitigate this, but we cannot guarantee bias-free output.

7. Your Rights Regarding AI

Under the Thailand Personal Data Protection Act (PDPA) and as a matter of our commitment to transparency, you have the following rights in relation to AI use:

• Right to know: You have the right to know when you are interacting with AI-generated content. Jenora clearly indicates AI-generated responses in the app interface.
• Right to opt out: You may disable cloud AI features from your account settings at any time. If you disable cloud AI, Jenora’s core self-guided tools remain available — including roadmaps, check-ins, journal tools, zone planning, worksheets, and progress tracking. AI-powered features will not be available, including companion chat, smart journal summaries, structured helpers, tone-matched paraphrasing, and AI-assisted pattern explanations.
• Right to contest: If you receive AI-generated content that you believe is inaccurate, harmful, or inappropriate, you can report it directly in the app or by contacting us at support@jenora.com.
• Right to human review: You may request that a human member of the Jenora team review a specific AI interaction you are concerned about. Contact us at support@jenora.com with the details.
• Right to withdraw consent: You may withdraw your consent to AI-powered personalisation at any time from within the app settings. This will stop your personal context data from being included in AI prompts.

8. AI and the Therapist Portal

Jenora’s AI features operate independently from the Therapist Portal. Specifically:

• AI-generated insights and companion responses are not displayed in the Therapist Portal. Therapists see only your raw in-app data (check-ins, mood logs, roadmap progress) — not AI interpretations of it.
• Your therapist’s notes and portal interactions do not influence the AI’s responses to you.
• AI does not have access to any communication between you and your therapist outside of Jenora.

9. AI Model Updates

We may update the AI models, providers, or features used within the Services over time. When we make changes that materially affect how AI interacts with your data, we will update this disclosure and notify you via in-app notice or email. The effective date at the top of this page reflects the most recent version of this disclosure.

10. Contact Us

If you have questions about how AI is used in Jenora, concerns about a specific AI response, or wish to exercise your rights related to AI, please contact us:

Neyda Technologies INC.
Operating as Jenora
Email: support@jenora.com