Futurion LLC, doing business as Alova
Last updated: October 13, 2026
Need help now? In the United States, call 988 or text 988 for the Suicide & Crisis Lifeline. You can also text HOME to 741741 for Crisis Text Line (crisistextline.org/text-us). In immediate danger, call 911. Do not wait for Alova to identify an emergency or contact someone for you.
Alova is an adult-only AI companion app, not a health-care, therapy, crisis, or emergency-response service. This protocol describes the automated safeguards and their limits. It is not a clinical assessment of any individual or a guarantee of a safe outcome.
Every companion is an AI character, not a human chat partner. Character emotions and relationships are simulated. Some messages and safety responses use text prepared by Alova. We explain the AI identity before companion interaction and provide ongoing reminders in the app.
Alova is for adults 18 and older. Companion chatbots may not be suitable for some minors, and Alova does not offer a minor-access companion mode. Credible underage information can lead to an eligibility restriction and review; it must not be treated as permission to continue adult romantic interaction.
For users with the required permissions, automated checks assess relevant message text for possible suicidal thinking, self-harm, imminent danger, disordered eating, and direct requests for crisis resources. Separate content and eligibility checks address prohibited sexual content and underage claims.
A direct crisis expression is treated differently from an ambiguous expression of ordinary distress. Ambiguous language can be recorded without showing a crisis card or changing the conversational response. Not every reference to a topic is interpreted as a statement that the user is in danger.
The system also distinguishes a general or ambiguous reference to disordered eating from a more actionable description of harmful behavior. An actionable situation can change the response to a non-diagnostic, non-shaming supportive approach and restrict inappropriate romantic or commercial behavior. An eating-related signal alone is not automatically classified as an emergency or assigned a 911 response. A separate crisis or imminent-danger indication can require a different response.
Detection can miss meaningful expressions, misread context, or respond unnecessarily. The absence of a card does not mean Alova has established that you are safe. We do not describe automated detection as a diagnosis or a clinically validated risk score.
When the system identifies a qualifying crisis expression or a direct request for crisis help, resources are presented through an app-controlled resource card or a compact resurfacing of an existing card. Resource text and contact information are not invented by the AI model.
The crisis presentation includes 988 and appropriate support resources. A presentation for apparent immediate danger prioritizes 911 with crisis support alongside it. Repeated qualifying expressions can surface resources again; reusing an existing card is not intended to hide access to help.
The companion is instructed to respond briefly and supportively, without describing self-harm methods, romanticizing harm, shaming the user, claiming to be a therapist, or pretending to be physically present. Supportive conversation is not a reason to delay reaching a trusted person, qualified professional, or emergency service. The app's resource presentation provides the contact details separately from the character's response.
Crisis-resource access is not a paid gift or relationship reward. Where an ordinary message or access limit would block a qualifying crisis interaction, the safety path provides an app-controlled response and resources rather than a purchase request. That does not guarantee a continuing live counseling conversation.
Before generation, model instructions prohibit encouragement or instructions for self-harm and prohibited sexual content. After generation, automated output checks look for defined harmful patterns. When a check identifies prohibited content, the response can be withheld and replaced with prepared text, with crisis resources added where the applicable crisis rule requires them.
Eating-disorder instruction or encouragement is also checked. A flagged response is replaced with a non-romanticized response; an eating-disorder output flag does not by itself assert that the user is in immediate danger.
These controls reduce risk but are not comprehensive understanding of every message. Unexpected outputs, missed detections, and technical failures remain possible. Report an unsafe response using the conversation's Report control or support@alova.dev. An error disclaimer does not remove Alova's legal responsibility to maintain appropriate safeguards.
Operational safety records can contain an event category, severity, time, identifiers linking the event to the account or conversation, and the response or resource-presentation category. Some crisis, age-claim, and output-moderation records can include a message or model-output excerpt of up to 200 characters. Other records, including ambiguous distress and eating-disorder output records under their excerpt rules, do not include such text. Safety information is not sent to ordinary product analytics as message text, memory text, or Safety Signals.
Authorized personnel may review limited information for a reported response, safety or eligibility matter, support request, security incident, or legal obligation. This is not continuous monitoring, an emergency alert system, or a promise that a person is reading while you chat.
Account deletion and relevant privacy requests erase or de-identify the covered records as described in the Privacy Policy and Consumer Health Data Privacy Policy. We may retain de-identified statistics only when the applicable legal standard is satisfied. Records that remain reasonably linkable to a person are not called anonymous merely because names were removed.
Safety processing can reveal consumer health data. It requires the permissions described in the separate Health Data Policy. It is part of the current companion-chat service and cannot be switched off while continuing the same chat feature. You may decline or withdraw permission in Profile → Account → Privacy, or contact privacy@alova.dev. Eligible browsing, account management, and privacy requests remain available.
Alova does not automatically transmit your conversation to 988, Crisis Text Line, law enforcement, or a nearby person. Following a crisis-service link, calling, or texting creates a direct interaction with that external service under its own practices. Limited legally required disclosures are addressed in our privacy policies.
No real-time human crisis team. There is no continuously staffed reviewer watching companion chats and no guarantee a safety event reaches a person promptly.
No emergency dispatch or reliable location. Alova cannot call emergency services on your behalf and does not use precise device location to locate you. Information you voluntarily share or technical network information is not an emergency-location capability.
No clinical validation claim. As of this revision, Alova has not completed an independent clinical review of this protocol. We do not claim that the safeguards are a medically validated treatment or risk-assessment system.
No substitute for real-world care. A companion is not a therapist, clinician, or crisis worker. You may leave a conversation at any time. Reaching out to a real person is compatible with using the app; Alova should not be relied on as your only source of support.
We review the public protocol when a release materially changes detection or response and at least quarterly. Changes must reflect the actual service; a planned safeguard is not described as already operating. Where legal reporting obligations apply, reporting uses the required privacy protections rather than publishing identifiable conversations or safety records.
Report an unsafe response or eligibility concern: support@alova.dev.
Questions about safety-data processing: privacy@alova.dev.
Support email is not an emergency channel. In immediate danger call 911; for crisis support call or text 988, or text HOME to 741741.