Safety

How Within responds when what you write suggests you might need support. Within is not a crisis service.

If you’re in danger right now: call 911 (US) or your local emergency number, or call or text 988 (US). All support options

What Within does

  • The Support button is on every screen. It works without AI, without a subscription, and even if you haven’t finished setting up.
  • When you save an entry, your browser can check its words for signs you might want support, and if so, show ways to reach people. This runs on your device, isn’t AI, and nothing about it is stored or sent anywhere. You can switch it off in Settings.
  • If what you wrote suggests you might be in danger, Within pauses reflection, connections and planning for that entry, and puts people first.
  • If words could mean several things, like a song lyric, a story, a quote, something from the past, or slang, Within asks briefly instead of assuming.
  • If an entry suggests immediate danger, Within shows emergency options straight away, with no questionnaire first.
  • AI output is checked before you see it. Anything that diagnoses, labels you, encourages harm, claims to be human, or tries to make you depend on it is thrown away.

What Within doesn’t do

  • It doesn’t score your risk or label you, and it can’t predict suicide or violence.
  • Nobody monitors entries, and it can’t send help or contact anyone for you, including your family.
  • It doesn’t promise secrecy or rescue, and it doesn’t pretend to be a person.
  • It can miss signs of danger, and it can misread what you meant. If you need help, reach out yourself; don’t wait for the app.

If AI or the safety checks are unavailable

You can still write, save and reach support. Within will say that reflection is unavailable. It won’t treat a failed check as a sign that you’re fine.

Your support circle

You choose who’s on it, and you see the exact message before anything opens. If you write about someone on your list in a way that suggests they might not be safe for you, Within won’t suggest them at that moment and will offer other options.

Review

The safety wording and routing follow published crisis-line guidance (911, the 988 Suicide & Crisis Lifeline) and haven’t yet been reviewed by an outside clinician. AI reflection stays switched off until that review is done.