Woosha includes optional features powered by artificial intelligence, specifically Anthropic’s Claude AI. These features are:
Journal Reflection Prompts are not an AI feature. The prompts are drawn from a fixed set written in advance and selected without any AI processing. Nothing you write is transmitted anywhere when you use them, and turning AI features off does not affect them.
AI features are OFF by default and require your explicit consent to enable. You can turn them on or off at any time in Settings, then AI Features. Turning them off does not affect other App features. The App is fully usable without AI.
AI can:
AI cannot:
AI-generated content may sometimes be incorrect, misleading, incomplete, or contrary to established clinical guidance. This is a known limitation of current AI technology.
Use your judgment. If an AI-generated insight feels wrong, it may be wrong. AI content is for personal reflection, not clinical guidance.
We review AI output quality periodically, but we cannot guarantee the accuracy of any AI-generated insight, and we do not review individual responses before you see them.
If you are experiencing a mental health crisis, do not rely on AI features for support. Contact emergency services or a crisis line immediately:
The App will not detect a crisis for you. If you are in distress, take direct action to get help from a human.
One narrow safeguard operates in the Socratic Questioning feature only. If the text you enter there contains wording associated with suicide or self-harm, the App does not send that text to Anthropic and instead shows pre-written crisis resource information. This is a simple keyword check, not a monitoring system. It does not apply to Weekly Insights, Thought Reframing, Journal Reflection Prompts, or Fear Ladder, it will miss many expressions of distress, and it is not a substitute for reaching a person who can help.
By enabling AI features in Woosha, you acknowledge and agree that:
What is sent to Anthropic is different for different features. We set it out feature by feature, because a general summary would not be accurate.
Weekly Insights
Sent: aggregated statistics drawn from your logs (counts, averages, correlations); structured fields from your panic events, including triggers, physical symptoms, emotional symptoms, coping methods, resolution, and any notes you have typed into those fields; counts of your reframing activity; technique effectiveness figures; and a short list of recurring keywords extracted from your journal entries.
Not sent: the full text of your journal entries, your name, your email address, your account number, or your authentication tokens.
Also sent: a pseudonymous identifier derived from your account by a one-way cryptographic hash. It is not your name or email, but it is stable between requests, so it is capable of linking your requests to one another. Under European data protection law this would be treated as pseudonymous rather than anonymous data.
Thought Reframing, Socratic Questioning, and Fear Ladder
Sent: the text you type, exactly as you wrote it. For reframing, this is the automatic thought you record and the evidence you write against it. For a fear ladder, it is the situation you describe. The thinking-error categories you select are also sent.
Not sent: any account identifier, your name, or your email address.
Important: we do not remove names, places, dates, or other details from this text before sending it. Whatever you write in these fields is what is transmitted. Please bear that in mind, particularly when writing about other people.
Predictions
Sent: aggregated pattern data, filtered by an automated step that attempts to remove name-like patterns, phone numbers, and email addresses before transmission, together with the pseudonymous identifier described above. This filter is a best-effort measure and will not catch every identifying detail.
Retention and training
Anthropic’s commercial API terms provide that inputs and outputs are retained for up to 30 days for trust and safety purposes and are then deleted, and that inputs and outputs are not used to train Anthropic’s models. We do not operate Anthropic’s systems, and this statement reflects Anthropic’s published terms as at the date of this disclosure rather than an assurance we are in a position to give ourselves.
Insights generated for you are shown only to you, stored in your account, and never shared with other users or third parties. You can delete them at any time.
If you turn off AI features, no data is sent to any AI service.
Never sent to Anthropic, in any feature:
Kept entirely on your device:
You are in control of AI features:
We currently use Anthropic (Claude API) exclusively for AI features. If we change AI providers in the future, we will:
Per Google Play requirements, we disclose that Woosha uses Generative AI to produce user-facing content in the features listed above. Woosha’s AI features:
If AI generates content that seems harmful, incorrect, or inappropriate:
Epic Media, LLC
Huntsville, Alabama, United States
Email: support@woosha.app
Privacy: privacy@woosha.app