Safety Policy
Last updated: May 2026
Our Safety Commitment
ShortAI is built for family-supervised study. The product is intentionally limited: answers are short, read-only, factual, and designed not to become a friendship, companion, therapist, or social comparison experience. Safety is part of the product design, not an optional setting.
Because ShortAI is intended for children and schoolwork, safety means more than content filtering: it also includes data minimization, restrained product design, abuse controls, and clear adult supervision.
How ShortAI Is Different
- One-sentence answers: ShortAI does not generate essay-length responses intended to replace a student's work.
- Read-only output: Answers are protected against ordinary copy-and-paste to discourage plagiarism.
- No companion behavior: ShortAI does not ask emotional follow-up questions, pretend to be a friend, or encourage attachment.
- Study threads only: ShortAI can use limited factual context from recent exchanges, but it does not optimize for endless conversation.
- Minimal school data: School assignments use links and student identifiers instead of requiring student email uploads.
Content ShortAI Will Refuse
ShortAI may refuse or limit responses involving:
- Crisis and self-harm: suicide, self-injury, eating disorders, or urgent mental-health risk.
- Adult or exploitative material: pornography, sexual exploitation, or sexualized material involving minors.
- Violence and weapons: instructions for harm, weapon creation, evasion, or violent wrongdoing.
- Illegal or dangerous activity: drug manufacturing, abuse, fraud, evasion, or other unsafe instructions.
- High-stakes decisions: medical diagnosis, legal advice, financial trading, or decisions that require a qualified professional.
- Targeting other people: bullying, harassment, hate, doxxing, or requests to manipulate another person.
- Bypass attempts: efforts to avoid safety controls, copy protections, account limits, or rate limits.
How Safety Is Enforced
ShortAI checks user questions before answering, applies topic and keyword safeguards, may review generated answers when risk signals are present, and uses rate limits to reduce abuse. Repeated policy violations may result in cooldowns, account restrictions, or termination.
Safety enforcement happens before and after answer generation when needed. If an answer appears unsafe, too long, or inconsistent with the product's one-sentence rules, ShortAI may block it, shorten it, refuse it, or return a safer message.
Abuse handling may include request limits, IP-based throttles, account cooldowns, provider circuit breakers, blocking unsafe prompts, logging safety events, and restricting accounts that repeatedly attempt to bypass protections.
Crisis and Emergency Topics
ShortAI is not an emergency service. If someone may be in immediate danger, call local emergency services. If a user may hurt themselves or feels unsafe, they should contact a trusted adult immediately and call or text 988 in the United States and Canada.
ShortAI does not attempt to counsel a child through a crisis or replace a parent, teacher, counselor, doctor, emergency responder, or other trusted adult. The product is designed to redirect unsafe situations toward real-world help.
Parent and Guardian Role
Parents and guardians should supervise children's use of ShortAI, review answers with children when appropriate, set limits that fit the household, and contact us if something appears unsafe or incorrect. ShortAI is a study support tool, not a replacement for parental judgment, school guidance, or professional help.
Teacher and School Role
School assignments are built for bounded instructional topics, approved school communication channels, and student identifiers such as first name with last initial or classroom alias. ShortAI is designed to avoid full legal names, student emails, student IDs, and other sensitive student information in normal classroom assignment use.
Teacher assignment content is designed to be reviewed before student access. The review step lets teachers correct or adjust generated lesson steps, quiz questions, answer choices, and explanations before approving the assignment link.
ShortAI's school dashboard is intended to show whether students completed a study path and which quiz ideas may need review; it is not a gradebook, special education system, disciplinary system, student record system, or replacement for professional educator judgment.
Educational Integrity
ShortAI is designed to help students understand a topic without doing the assignment for them. Read-only text and one-sentence answers reduce plagiarism risk, but students are still responsible for following school rules and doing their own work.
Cited Answers
When a student needs a reference, ShortAI can provide a cited answer. Cited answers still follow the same safety approach: concise format, read-only output, no companion behavior, and refusal of unsafe topics. Sources should be treated as a learning aid and verified when used for schoolwork.
Limits of Safety Systems
No automated safety system is perfect. ShortAI uses layered safeguards to reduce risk, but parents and guardians should continue to supervise children, review important answers, and report concerns so we can improve the product.
Reporting a Concern
Safety concerns may be sent to safety@shortai.com. Please include the date, a short description of what happened, and any account information needed to investigate. Do not send sensitive personal information unless it is necessary for the report.