Murph Health AI Safety Disclosure

Effective Date: April 29, 2026

Last Updated: July 23, 2026

Murph is an artificial intelligence system. Messages, summaries, calls, recommendations, and other outputs or actions may be generated automatically and may not be reviewed by a human before they are shown, sent, or used in a workflow.

Murph uses AI-assisted systems to help organize personal wellness information, summarize records, compare protocols, draft reflections, retrieve sources, parse files, support bounded self-experiments, and carry out user-directed tasks. This disclosure explains the safety boundaries for those AI features.

1. Murph is not medical care

Murph is a personal wellness, journaling, education, and self-experimentation tool. Murph is not a doctor, clinician, hospital, pharmacy, therapist, dietitian, trainer, emergency service, medical device, remote patient monitoring service, or clinical decision support system.

AI-generated output from Murph is not medical advice and is not a substitute for professional medical judgment. Do not use Murph to diagnose, treat, cure, prevent, monitor, or alleviate any disease, injury, disability, or medical condition.

2. AI output can be wrong

Murph output may be inaccurate, incomplete, biased, stale, misleading, unsafe, or not applicable to your situation. Output may reflect errors in your inputs, wearable data, imported records, lab values, files, third-party APIs, search results, source summaries, retrieval systems, or model behavior.

Use your own judgment. Consult a qualified clinician before making decisions about medications, supplements, treatment, diagnostic testing, medical devices, exercise, sauna, cold exposure, fasting, alcohol changes, sleep interventions, or other health-related activities.

3. Patterns are not proof of causation

Murph may compare time periods, identify associations, summarize trends, and generate hypotheses from your data. A before-and-after change, correlation, timing relationship, score, label, or suggested explanation does not prove that an intervention caused an outcome, that the outcome will recur, or that an intervention is safe or appropriate for you.

Self-experiments may be affected by normal variation, measurement error, missing data, device limitations, illness, travel, stress, sleep, medication, nutrition, training load, menstrual cycle, environmental conditions, and other confounding factors.

4. Do not use Murph for emergencies

Do not use Murph for emergencies, urgent care, crisis response, acute symptoms, medication emergencies, overdose, poisoning, chest pain, difficulty breathing, fainting, severe allergic reactions, suicidal thoughts, self-harm, disordered-eating crises, or other time-sensitive clinical decisions.

If you think you may have a medical or mental-health emergency, contact emergency services or seek immediate in-person care.

5. No medication or clinical decision authority

Do not use Murph to determine whether to start, stop, change, time, increase, decrease, combine, or dose any medication, supplement, therapy, treatment, diagnostic test, medical device, or clinical care plan.

Murph may not detect dangerous interactions, contraindications, allergies, adverse effects, abnormal results, or other safety issues. Murph does not continuously monitor your data or guarantee that it will warn you about concerning symptoms, metrics, or trends.

6. Higher-risk users should get professional guidance

If you are or may be pregnant, nursing, trying to conceive, under medical care, taking medication, have a history of an eating disorder, or have a heart, metabolic, neurological, psychiatric, respiratory, immune, kidney, liver, endocrine, sleep, or other medical condition, consult a qualified clinician before using Murph or following any protocol.

Stop any activity and seek medical care if you experience concerning symptoms.

7. AI providers and health data

Murph may use model, search, parsing, transcription, embedding, retrieval, voice, browser, and automation providers to provide requested features. Murph tries to limit data shared with those providers to what is reasonably necessary for the requested feature.

Murph does not use health data, consumer health data, HealthKit data, wearable data, journal content, health files, health memories, health prompts, assistant outputs based on health data, or derived health insights to train, fine-tune, or improve Murph's or any third party's general-purpose AI models.

Murph does not route Murph-managed health data to third-party model providers unless the relevant contract, service setting, or deployment control prohibits model training on Murph health data.

Some Connected Services do not permit their data to be supplied to or used in the operation of AI systems, or permit AI use only through a provider-controlled interface or separate written authorization. Connecting a source does not override those restrictions. Data from such a source may be unavailable to Murph's AI features, persistent memories, embeddings, retrieval, cross-source analysis, experiments, group features, or newsletters. Murph may disable or limit an integration when a permitted processing path is unavailable.

8. AI calls and user-directed actions

At your request, Murph may use an automated or AI-generated voice, send messages, use browser tools, make bookings or purchases, submit information, or take other actions. These actions may fail, be delayed, be sent to the wrong recipient if information is incorrect, execute differently than intended, or not execute at all.

Murph has only the task-specific authority you give it. Murph is not your general agent, healthcare proxy, attorney-in-fact, clinician, or fiduciary. Murph may not consent to treatment, modify a prescription, authorize emergency care, or make another clinical decision on your behalf.

Review material terms, recipients, information being transmitted, and confirmations before relying on an action. Third parties may independently record, retain, or process communications and transaction information under their own terms and privacy notices.

9. Human review and support

Murph personnel do not review the contents of your hosted health data except where needed to provide support you request, investigate security or abuse issues, debug service problems, comply with law, enforce our Terms, or protect the rights, safety, privacy, or property of Murph, users, or others.

Hosted Murph is not zero-knowledge, operator-blind, or end-to-end encrypted unless Murph expressly says otherwise for a specific feature. Hosted Murph may need to decrypt content to run requested tasks, operate hosted features, troubleshoot user-reported issues, investigate abuse, or comply with law.

10. Your control

You can choose whether to use optional integrations and optional AI-powered features. Some features may require authorization, permission, or consent to process Consumer Health Data or use specific providers. You may disconnect optional integrations, withdraw optional feature permissions for future processing where supported or required by law, and request deletion through available privacy controls.