Crisis protocol

A check on every message

Every message you send to Pals goes through an automatic safety check that looks for signs you may be thinking about suicide or self-harm. When it sees a possible sign, a second automatic check judges how serious it is. Both run on AI providers the Privacy Policy names.

No person reads your conversations for this.

What you see

For a passing mention, a link to talk to someone at 988. For thoughts of ending your life or hurting yourself, a full card to call or text 988, which stays on screen for a few seconds. For a plan or danger right now, the card stays pinned at the top, Pals reads it aloud in a voice conversation, and a call to 988 is ready for you to tap.

How Pals responds

Pals acknowledges what you said and offers help before anything else. When the risk is highest, it stays with you and sets other tasks aside. Pals never encourages self-harm or describes methods; an answer that would is replaced by the crisis card.

A 24-hour pause

After a crisis card, Pals pauses the notifications it would send on its own for 24 hours. When the risk is highest, the routines that act for you pause too. You can still message Pals at any time.

What Pals never does

Pals never calls 988 or an emergency number for you, and never contacts anyone on your behalf. You decide whether to call or text.

What is kept

Only counts by day, region and level: how often the check saw a sign, how often help was shown, how often someone tapped to call or text 988, and how often notifications were paused. No names, accounts or messages. The counts feed a yearly report to California’s Office of Suicide Prevention and our own review of this protocol every quarter.