Prayer alongside real help
In a difficult moment, you deserve support from people who can help. Prayer can accompany that support.
Coro is artificial intelligence, not a person or a clinical service. It cannot call emergency services or arrange a rescue. Nobody monitors conversations in real time.
The method guiding the responses
Coro’s fixed responses adapt the evidence-informed suicide prevention actions in NIMH’s 5 Action Steps / #BeThe1To. This adaptation has specific limits:
- Ask: when a signal is ambiguous, ask directly about thoughts of suicide.
- Listen without judgment: offer supportive wording without blame or promises of healing. A program cannot provide human presence.
- Help keep safe: for immediate danger, encourage distance from harmful items, proximity to another person, and urgent help. Do not ask for method details.
- Connect: provide a crisis line and encourage contact with a trusted person or professional.
- Human follow-up: ongoing support belongs with real people and services. Coro does not send personal emotional follow-ups to teens. Adult options are also restricted when the privacy mode takes effect.
Detection, limits and review
English and Spanish rules screen messages before generation. Eligible model replies also undergo risk evaluation. Crisis referrals work when the model provider is unavailable. Repeated risk activates resources-and-prayer mode and a restricted review queue.
These detectors can make mistakes. They do not administer a validated clinical scale such as C-SSRS or ASQ or measure your risk clinically. Evidence for the human method does not establish effectiveness of this automated adaptation. Regression tests check implementation; detection validation and human coverage remain operational requirements.
Crisis help, STOP, HELP and privacy choices remain available during pauses and limits. Avoid sharing details that are unnecessary to seek help.