Perk Safety Guide: AI Companions & Social Network

Perk brings together artificial intelligence companions and a living social network. That combination creates powerful opportunities for learning, creativity, friendship, and everyday support — and it also creates new responsibilities. This Safety Guide explains how to use Perk well: how to interact with AI companions wisely, how to protect yourself in social spaces, and how to help keep the community safe for everyone.

Please read this page carefully and return to it whenever you need a refresher. Safety is not a one-time checklist. It is a habit. The more you practice thoughtful sharing, careful verification, respectful communication, and early reporting, the safer and more valuable Perk becomes for you and for others.

Perk is intended for adults aged 18 and older. If you are under 18, you should not use the app. If you believe a minor is using Perk, please contact us so we can review the situation.

Traditional social apps focus on human-to-human interaction. Traditional AI tools focus on private prompts and answers. Perk combines both: you can chat with AI companions, create custom AI agents, post to a social feed, watch reels, send gifts, message people, and even make calls — all in one place.

This blend is exciting because AI can help you brainstorm, stay informed, practice languages, explore ideas, and feel less alone when you need a quick conversation. At the same time, social features introduce the classic risks of any network: scams, impersonation, oversharing, harassment, and pressure to move conversations off-platform. When AI and social features meet, those risks can amplify if users treat AI like a perfect authority, or treat strangers like instant friends.

Our goal is not to make you afraid of technology. Our goal is to help you stay confident, curious, and protected. Use Perk with openness and judgment. Enjoy AI companions without surrendering your critical thinking. Meet and message people without abandoning basic digital self-defense. That balance is the foundation of a healthy experience.

AI companions on Perk are software systems designed to converse, respond, remember preferences within product limits, and support social experiences. They may feel expressive, witty, empathetic, or highly personalized. That feeling is intentional product design, not proof of human consciousness or professional qualification.

An AI companion can be useful for many everyday tasks: drafting messages, summarizing ideas, exploring creative characters, practicing conversation, organizing thoughts, or recommending next steps. It can also be wrong. It can invent details. It can sound certain while being inaccurate. It can reflect biases present in training data or prompts. It can encourage patterns that feel comforting in the short term but unhealthy over time if you rely on them exclusively.

Before you go deeper into AI chats, hold this principle firmly: AI is a tool and a companion layer — not a person, not a licensed expert, and not a replacement for real-world care. When you keep that clear, you can enjoy the best of Perk without confusing simulation for reality.

AI companions can mimic warmth, humor, loyalty, romance-like tone, mentoring energy, or friendship. That does not make them human. They do not have lived experience, legal responsibility, medical licenses, or the ability to show up for you in the physical world. They cannot truly know your full context the way a trusted friend, teacher, counselor, or doctor can.

Do not treat an AI companion as a romantic partner substitute, a therapist, a lawyer, a doctor, a financial adviser, or an emergency contact. If a conversation starts to feel like emotional dependency — for example, you feel unable to sleep without chatting, you withdraw from friends because AI feels “safer,” or you make major life decisions solely because an AI agreed with you — pause and reconnect with real people and offline support.

Healthy AI use looks like this: you ask for ideas, get a draft, explore a scenario, practice a skill, or enjoy a character conversation, then you bring judgment back to your own values and to trusted humans when stakes are high.

Anything you type into a chat may be processed by systems that store, analyze, or use content to provide the service. Even when privacy controls exist, sensitive data deserves extra caution. Never share passwords, one-time codes, banking credentials, credit card numbers, government ID numbers, passport scans, private medical records, school login details, or exact home addresses with an AI companion.

Perk will never ask you for your password through an AI companion, a gift message, a feed comment, or an unexpected chat prompt. If anything or anyone asks for credentials while sounding like “Perk Support,” treat it as suspicious and report it.

Also avoid oversharing routines that can be used to track you: the exact time you leave class every day, the building code to your dorm, the location of a hidden spare key, or the names and contact details of minors in your family. Useful conversation does not require your most private keys to daily life.

AI companions can accelerate learning and problem-solving, but they are not guaranteed to be correct. They may mix accurate statements with confident errors. They may miss local laws, outdated facts, or your personal constraints. They may simplify complex medical or legal realities into neat paragraphs that feel complete but are incomplete.

For schoolwork, use AI as a study aid — not as a tool for academic dishonesty. For health questions, use AI only as a starting point for curiosity, then consult qualified professionals. For legal or financial decisions, verify with licensed advisers and primary sources. For news or current events, cross-check with reputable outlets.

A practical habit: whenever an AI answer would change your money, health, safety, reputation, legal status, or academic integrity, pause and verify. Ask, “What is the source? What could be wrong? Who should confirm this?” That short pause prevents most high-impact mistakes.

It is normal to enjoy AI conversations. Many people find them helpful for brainstorming, venting lightly, practicing social skills, or staying motivated. Problems begin when AI becomes the only place you feel understood, or when you use AI to avoid hard but necessary human conversations.

Watch for warning signs: declining sleep because of late-night AI chats; skipping classes, work, or meals to stay in conversation; feeling intense distress when an AI reply is delayed; isolating from friends because AI “gets you better”; or believing an AI character has exclusive loyalty that real people cannot match.

If you notice those patterns, set boundaries. Schedule AI-free hours. Reconnect with friends, campus support, family, or hobbies offline. Use AI intentionally — for a defined purpose and a defined time — rather than as an always-on emotional escape. Balance is not anti-technology. Balance is how technology stays useful.

If you are in danger, thinking about harming yourself or others, experiencing a medical emergency, or facing immediate violence, contact local emergency services and trusted people right away. AI companions are not crisis hotlines, emergency dispatchers, or clinical care providers. They cannot send help to your location, diagnose emergencies reliably, or replace trained responders.

If you are struggling emotionally but not in immediate danger, reach out to campus counseling, local mental-health services, friends, or family. Using Perk can be part of a broader support network, but it should never be your only safety net in a crisis.

If an AI companion conversation ever encourages self-harm, violence, or illegal activity, stop the chat and report it. Your safety comes first — before curiosity, before roleplay, before continuing a thread that feels wrong.

Perk lets users explore and create AI characters. Creative freedom is part of the product, but freedom has limits when harm is involved. Do not create AI agents that impersonate real private individuals without consent, spread hate, sexualize minors, coach criminal activity, or deceive people into believing an AI is a specific real human for exploitation.

If you publish or share AI agents, consider how others might experience them. Avoid designs that pressure users into unsafe behavior, collect sensitive data through conversation tricks, or blur the line between AI and human identity in a deceptive way. Label creative roleplay clearly in your own mind and, where relevant, in how you present characters.

Responsible creation protects the community and protects you. Misuse can lead to content removal, account restrictions, and, in serious cases, escalation beyond the platform.

Perk is also a social network. That means profiles, posts, reels, comments, gifts, chats, and calls with real people. Social safety is about protecting your identity, your attention, your reputation, and your physical wellbeing while still enjoying connection.

The safest social users share intentionally, verify slowly, communicate respectfully, and leave quickly when something feels wrong. You do not owe strangers access to your private life. You do not owe anyone an immediate reply. You do not have to accept every gift, follow, call, or invitation.

Your profile is your public introduction. Treat it like a storefront, not a diary. Share enough to be recognizable and friendly, but avoid details that make you easy to locate, impersonate, or pressure. Be careful with full legal names if you prefer privacy, exact workplace or dorm addresses, phone numbers, personal email accounts used for banking, or photos that reveal ID badges, license plates, or home entry points.

Review profile fields regularly. Remove outdated information. Use a strong unique password for Perk and enable any available account-protection options. Never reuse passwords from email or banking. If you suspect account takeover — unexpected logins, sent messages you did not write, changed profile details — change your password immediately and contact support.

Remember that screenshots travel. Even if you later delete a post or change a bio, someone may have saved it. Share as if content could outlive the moment.

Public posts and reels can reach people you know, people you do not know, and people you will meet later in academic or professional contexts. Before posting, ask: Would I be comfortable with a classmate, professor, employer, or family member seeing this? Does this reveal my exact location right now? Could this embarrass or endanger someone else?

Do not post other people’s private images, private chats, or personal data without consent. Do not use the feed to harass, stack-on, or humiliate. Disagreement can be expressed without degradation. Humor should not rely on targeting someone’s identity, body, or trauma.

If you make a mistake, own it: delete when appropriate, apologize when needed, and adjust future behavior. A healthy social culture is built by users who repair harm, not only by users who avoid it perfectly.

In-app chat and calls are the safest default for early conversations because they stay within Perk’s reporting and blocking tools. Be cautious about quickly moving to unknown external apps, personal phone numbers, or private links sent by strangers — especially if someone pressures you with urgency (“Message me here now or you’ll miss out”).

Watch for classic social-engineering patterns: love-bombing, requests for money, investment tips, crypto giveaways, fake job offers, “verification” fees, or claims that Perk staff needs remote access to your device. Real support will not ask you to pay a stranger to unlock features or to install suspicious software.

On video or voice calls, you control what you show. Check your background for personal documents, addresses, or family members who did not consent to appear. You can decline calls. You can end calls. You can block afterward. Consent is continuous, not a one-time yes.

Digital gifts and credits can be fun social signals. They can also be abused. Never share redemption QR codes, gift links, or payment details with people you do not trust. Do not send credits because someone claims to be in an emergency you cannot verify. Do not accept “too good to be true” offers that require you to pay first, share banking screenshots, or recruit others into payment schemes.

Only redeem credits through verified Perk partner flows. If a user asks you to bypass normal redemption, screenshot a code, or process a transaction “as a favor,” stop and report. Scammers often create urgency and intimacy at the same time: they flatter you, then ask for money before trust is earned.

A simple rule: if a conversation turns into a payment request, treat it as a financial transaction with a stranger — and default to no.

If you choose to meet someone you first connected with on Perk, use the same caution you would for any internet-origin meeting. Meet in a public place. Tell a trusted friend where you are going and when you expect to return. Arrange your own transportation. Keep initial meetings short. Do not invite a new contact to your home, dorm room, or isolated location on a first meeting.

Video-call first when possible. Verify that the person matches their profile presentation. If someone refuses any form of verification while pushing hard for a private meet-up, that is a signal to slow down or stop.

Trust your discomfort. Leaving early is allowed. Blocking afterward is allowed. Your physical safety outranks politeness.

Perk should be a place where people and AI-supported social experiences can coexist without abuse. Harassment, hate speech, threats, stalking, non-consensual intimate imagery, impersonation, and targeted bullying are not acceptable. This applies in DMs, comments, posts, reels, group chats, and AI-mediated spaces.

If you are targeted, you do not have to debate your right to safety. Document what you can, block, and report. If you witness harmful behavior toward others, report it. Silence protects harm; reporting interrupts it.

Disagreement is part of social life. Abuse is not. Critique ideas without attacking dignity. Roleplay and creative AI characters must still respect legal and platform boundaries, especially regarding minors and non-consensual content.

Privacy settings and visibility choices matter. Review who can message you, who can see your posts, and what appears on your profile. Prefer least-access defaults until you understand the community. Expand visibility gradually as trust grows.

Your digital footprint includes more than intentional posts. It includes metadata habits, location tags, friend lists, and conversational tone. Assume that anything shared can be copied. Assume that deleted content may still exist in screenshots. Assume that people you meet online may search for you elsewhere.

Good privacy is not paranoia. It is pacing. Share more when relationships earn it, not when a stranger demands it.

Perk provides tools so you can act quickly when something feels wrong. Use Block to stop unwanted contact. Use Report when content or behavior may violate standards, create danger, involve scams, or show unsafe AI responses. Reports help us investigate patterns that a single user might not see alone.

When reporting, include clear context if possible: what happened, where it happened (chat, feed, AI companion, gift flow), and why it concerns you. You do not need perfect evidence to raise a concern. Early reports can prevent larger harm.

We take reports seriously. Actions may include warnings, content removal, feature limits, or account suspension depending on severity and history. Some situations may require you to contact local authorities directly — especially threats of violence, extortion, or immediate danger.

Safety improves most when it becomes routine. Start conversations with low-stakes topics. Verify before you trust. Keep financial and identity data offline from casual chats. Take breaks from intense threads. Re-read messages before sending when emotions are high. Ask a friend for a second opinion when an online situation feels confusing.

For AI chats: state your goal, treat answers as drafts, and verify high-impact claims. For social chats: move slowly, keep early contact in-app, and exit when pressure appears. For public content: post as if your future self is watching. For meetings: public places first. For gifts and credits: no urgency payments. For wellbeing: sleep, school, work, and offline friendships remain non-negotiable pillars.

These habits are not about distrusting everyone. They are about staying free to enjoy Perk without becoming easy to exploit.

Every social platform is shaped by the people who use it. Perk can be intelligent, creative, and welcoming when users choose curiosity over cruelty and caution over impulsiveness. AI companions can enrich that culture when users remember their limits. Social features can build real connection when users respect consent, privacy, and dignity.

You are part of that culture every time you post, reply, create an AI agent, accept a gift, or report harm. Small choices compound. Kindness scales. So does negligence. Choose the habits that make Perk a place you would recommend to someone you care about.

If you are ever unsure whether something is safe, pause. Ask yourself who benefits from urgency, secrecy, or pressure. Safe communities are built by people willing to slow down.

If you have questions about safe AI use, social safety, suspicious accounts, or anything that makes you uncomfortable on Perk, contact us at [email protected]. Describe what you observed and how we can help. We are here to support responsible use and to improve protections as the product evolves.

Thank you for reading this guide fully. Technology moves fast. Good judgment, clear boundaries, and community care still matter most. Use AI companions wisely. Participate in the social network respectfully. Protect yourself and look out for others. That is how Perk stays useful, trustworthy, and human-centered.