AI Policy
Effective date: August 5, 2026
Version: 2
This Policy explains where Bornly uses artificial intelligence, what data is involved, what controls apply, and what AI must not be used for. It supplements the Privacy Policy, Terms, and Medical Disclaimer.
1. Provider and model role
Bornly currently uses OpenAI API models for the production uses listed below. Bornly selects the feature purpose, input, instructions, access rules, and output shown or stored. OpenAI processes requests as Bornly's service provider. The exact model may change for quality, safety, cost, or availability without changing the purpose described here.
Bornly does not authorise OpenAI to train general-purpose models on Bornly API inputs or outputs. OpenAI's current default API controls do not use content for training and may retain abuse-monitoring content for up to 30 days unless an approved shorter-retention control applies or law requires longer.
2. User-facing AI features
| Feature | Purpose | Typical inputs | Output and human role |
|---|---|---|---|
| Gutly and Crunchly meal AI | Identify foods and estimate a user-requested meal entry | Meal descriptions/photos and an AI-created food log plus edit instructions | Suggested food and nutrition data for user review |
| Gutly and Crunchly food-catalogue contribution AI | Extract product-label information and validate or cross-check a user-requested public contribution or correction | Product-front, barcode, ingredient-list and nutrition-label images; typed submissions/corrections, ingredients, nutrients, serving information and relevant catalogue context | Suggested catalogue data for user or staff review |
| Gutly summaries and Prof Gut chat, where available | Generate journal summaries or answer user questions about logged gut-health and lifestyle information | Account or display name where included by the feature; the question or requested reporting period; relevant food, symptom, bowel, mood, body, profile, preference, and journal context | AI-generated wellness information for user review; not medical advice or clinical monitoring |
| CrunchBuddy, where available or shown as a preview | Provide nutrition-related conversation or guidance | User messages and relevant food, nutrition, goal, profile, or conversation context | AI-generated information for user review; not medical advice |
| Baby Names suggestions | Suggest names from stated preferences | Liked and excluded names, gender filter, and suggestion-refinement history | Suggested names and explanations |
| Baby Names description search | Interpret a description and find matching names | Description, gender filter, liked and excluded names, and up to 12 prior messages from the same suggestion conversation | Suggested names for user review |
| Baby Names full-name analysis | Analyse the exact name combination requested | First, optional middle, and last name entered by the user | Subjective score and explanation |
| Night Lab reports, where available | Draft a report about a selected sleep experiment or journal period | Authorised sleep or HealthKit records, self-reports, experiment settings, adherence, notes, and derived metrics | AI-generated wellness report for user review; not diagnosis or monitoring |
Availability varies by app, version, territory, subscription, rollout, and configuration. Some experiences may be previewed, paused, changed, or discontinued. Earlier output may remain subject to the retention and deletion limitations described in the Privacy Policy.
Bornly does not deliberately add the account name, email address, or internal user ID to food-AI or Baby Names suggestion requests unless a feature needs content that the user entered or operational context linked to the request. Gutly AI context can include an account or display name. Free text, images, names, or combinations of data can still identify a person. Full-name analysis necessarily sends the name entered. This Policy describes the material inputs and purpose even where a shorter in-app notice does not repeat every detail.
3. Background and internal AI
Bornly also uses AI in limited operational workflows:
- Feedback and support triage: when the submitter permits the offered AI use, Bornly may send the submitted text or attachment, app/device/account or subscription context, diagnostics, and excerpts or patterns from similar reports. AI can categorise, summarise, match support material, or draft a reply or engineering issue. Staff may place submitted feedback in restricted support, Slack, or GitHub workflows.
- Food-catalogue operations: AI may translate, classify, enrich, validate, or generate food content and imagery. These workflows generally use product or catalogue data, but a user submission can contain personal or health-related material.
- Questionnaire drafting and translation: authorised staff may use AI to create draft questions or translations from a staff brief, product materials, and approved survey content. Machine-translated text may be displayed without review in every language and can contain errors.
- Survey and research analysis: AI may receive question text, answers including free text, response context, identifiers, locale or timestamps, and aggregate results to prepare internal summaries or research material. Participation is voluntary where presented as a survey, but submitted answers can still contain sensitive data.
- Marketing and product drafts: authorised staff may use AI to create internal audience, product, or ad-copy drafts from app materials, business context, research, survey content, and results. A human remains responsible for claims and publication.
Internal AI must not be used to infer a user's medical condition or target advertising based on health data. A human remains responsible for publishing surveys, claims, marketing, external replies, or other consequential actions.
4. Consent and controls
Some AI features are initiated when the user deliberately selects a labelled AI action or submits content to it. Where Bornly relies on consent, including explicit consent for sensitive data, the feature must first request a valid choice for the stated purpose and material data categories. A feature may provide a shorter point-of-use notice; this Policy provides the fuller description. Feedback AI uses the permission shown in the feedback flow where offered. Survey participation is voluntary where presented as a survey, but participation alone is not treated as explicit consent for a separate AI purpose.
Consent or another affirmative choice can be withdrawn through an available app control or privacy@bornly.com. Withdrawal blocks future processing that relies on that choice but does not itself delete source records, prompts, earlier output, survey answers, or evidence of the choice. Use an available deletion control or contact Bornly to request deletion.
5. Data minimisation and safeguards
Bornly applies measures such as feature entitlements, authentication, rate limits, prompt limits, access controls, logging, and input minimisation according to the feature and risk. Bornly may store prompts and outputs, including raw food-AI prompts and responses, for feature history, quality and accuracy review, abuse prevention, support, or incident investigation. Those records may contain identifying or sensitive information that a user entered even when no user ID is deliberately added.
Users should not enter another person's sensitive data or identifiers without authority. Avoid information that is not needed for the request.
6. Transparency and output labelling
Bornly identifies AI through this Policy and, depending on the feature, through names, consent screens, buttons, loading text, reports, or surrounding explanation. AI output is not human-authored or clinically reviewed unless expressly stated. Bornly may add or improve labels as laws and interfaces change.
7. Prohibited reliance and use
Do not use AI output to:
- diagnose, treat, monitor, or prevent a condition;
- select or change medication, supplements, treatment, or a prescribed diet;
- decide whether urgent or emergency care is needed;
- assess fetal or maternal well-being;
- make an employment, credit, insurance, legal, housing, education, or similarly consequential decision; or
- make a consequential decision about another person.
Bornly does not offer AI features for medical-device or healthcare-professional use. Output can be inaccurate, incomplete, biased, or fabricated. Verify important information with authoritative sources and qualified people.
8. Testing, review, and incidents
Bornly reviews AI features, prompts, output formats, access controls, and safety language in proportion to the feature and risk. We monitor quality, privacy, safety incidents, harmful output patterns, and complaints and may restrict, change, or disable a feature in response.
9. Rights and questions
Privacy rights, retention, recipients, and international transfers are described in the Privacy Policy. Questions or complaints may be sent to privacy@bornly.com.
We will update this Policy when AI purposes, providers, or material data flows change and will request new consent where required.