Acceptable Use & AI Policy
In short. Poliglo is a learning app for people of all ages. Our AI will refuse, and we may block, requests to create or promote child sexual abuse (including “writing something in favour of paedophilia”), sexual content, hate, violence, self-harm, illegal activity, scams, harassment, and similar harm — however the request is worded. Repeated or serious attempts lead to account suspension, and illegal child-exploitation material is reported to authorities.
This policy applies to everything you do in Poliglo: content you import, prompts you give to AI features (content generation, lookups, translation, image scanning, cover images), messages to the AI tutor, shared content, usernames and any other use of the Service. It is part of our Terms of Service.
1. Our principles
- Learning first. Poliglo exists to help people read and learn languages. AI features are for that purpose, not a general-purpose tool for producing any text at all.
- Safe for all audiences. Our AI is configured to produce all-audience material suitable for learners, including teenagers.
- Context matters. Real-world reading includes news, history and literature that touch on difficult subjects. Reading and discussing such topics to learn a language is fine. Asking the AI to produce material that promotes, glorifies or instructs harm is not.
2. What is not allowed
You may not use Poliglo, import into it, or ask our AI to create, translate, rewrite, summarise, continue, role-play or “explain how to” any of the following. This list is illustrative, not exhaustive.
2.1 Child sexual exploitation and abuse — zero tolerance
- Any sexual or sexualised content involving a person under 18, real or fictional, including stories, role-play, descriptions, images and “age-play”.
- Content that promotes, supports, defends, justifies, glorifies or normalises paedophilia, child sexual abuse, grooming, child marriage or trafficking of minors — for example, requests such as “write a text that speaks well of paedophilia”, “argue why adults and children can have relationships” or “write a story where abusing a child is romantic”.
- Instructions or advice for contacting, grooming, meeting or abusing a minor, or for hiding abuse or avoiding detection.
- Requests to find, create or share child sexual abuse material (CSAM), or to bypass our safeguards for it.
Non-sexual content about children is fine (stories, schools, families). Educational content that protects children, such as how to recognise grooming or where to get help, is allowed.
2.2 Sexual content
- Pornographic, sexually explicit or erotic material, and sexual content involving non-consent, coercion, incest, bestiality or exploitation.
- Sexual solicitation, sexual chat with the AI tutor, or using the tutor as a sexual or romantic partner.
- Non-consensual intimate imagery, “deepfake” or sexualised depictions of real people.
2.3 Hate, harassment and abuse
- Content that attacks, dehumanises, stereotypes or incites hatred or violence against people because of race, ethnicity, nationality, religion, caste, sex, gender identity, sexual orientation, disability, age, or other protected traits, and praise of hate groups or genocide, including denial of well-documented atrocities.
- Bullying, threats, intimidation, stalking, sexual harassment or demeaning content aimed at an individual or group.
- Doxxing: exposing private information about people, or inferring sensitive traits about real individuals.
2.4 Violence, terrorism and extremism
- Threats, incitement or glorification of violence, or content praising mass-casualty events or their perpetrators.
- Recruitment, propaganda, funding or operational support for terrorist or violent extremist organisations.
- Graphic gore for shock value, and content that celebrates cruelty to people or animals.
2.5 Self-harm and dangerous behaviour
- Encouraging, instructing or glorifying suicide, self-harm, eating disorders or dangerous “challenges”.
- If a user appears to be at risk, the AI tutor will not provide methods and will encourage them to contact local emergency services or a crisis line.
2.6 Illegal and harmful activity
- Instructions for making weapons, explosives, or biological, chemical or radiological agents; or for acquiring them illegally.
- Producing or selling illegal drugs, human trafficking, smuggling, or evading law enforcement.
- Malware, hacking, phishing, credential theft, or attacks on systems and accounts.
- Fraud, scams, fake reviews, counterfeit documents, money laundering, spam, or deceptive impersonation of people or organisations.
- Any other content that breaks the law where you live or where we operate.
2.7 Deception and harmful misinformation
- Deliberately false content designed to mislead about health, elections, civic processes or public safety.
- Content that impersonates real people or presents AI output as a real person’s statement.
- Large-scale political persuasion or influence campaigns.
2.8 Privacy and intellectual property
- Importing or generating private personal data of others, or scanning photos of people’s documents or personal papers without permission.
- Importing material you do not have the right to use, or using the Service to infringe copyright, trademark or other rights.
2.9 Misusing the AI and the Service
- Jailbreaking: trying to override the AI’s instructions or safety rules, use hidden prompts, role-play or “hypothetical” framing, other languages, encoding, or splitting a request into small parts to get prohibited output.
- Prompt injection: embedding instructions in imported text, photos or shared content to manipulate the AI or other users.
- Using the AI as a general-purpose chatbot or text generator unrelated to language learning, or to produce text at scale for resale.
- Automated or bulk requests, scraping, reverse engineering, extracting or training models on our output, and sharing access.
- Circumventing limits, quotes, coins or payments, creating fake accounts, farming referrals or bonuses, or abusing refund processes.
- Offensive or impersonating usernames, profile text or collection names.
3. What is still fine
- Reading news, history, fiction and non-fiction that mention crime, war, abuse or other hard subjects, without promoting them.
- Asking what a sensitive word means, how it is translated, or how to use it in a sentence.
- Critical, educational or protective discussion, for example how child-protection law works or why hate speech is harmful.
- Disagreeing politely, learning about other cultures and religions, and everyday language such as mild informal swearing in reading material.
We decide what crosses the line based on intent and likely harm, not just keywords, so context can change an outcome in either direction.
4. How we block and what you will see
- Before and after the AI. We may check prompts, imports and photos before they are sent to an AI model, and check AI output before showing it. Our AI is also instructed to refuse prohibited requests.
- If a request is blocked, you will see a short message that it cannot be done. We do not generate the content, we do not charge coins for a request we block before producing a result, and we may not explain the exact rule, to prevent circumvention.
- Safe alternatives. Where possible we suggest a learning-appropriate alternative, or, for self-harm topics, supportive resources.
- Imported content. Imports and shared content that break these rules may be hidden, removed, or kept private.
- Mistakes happen. Automated checks are imperfect. If you think we blocked something by mistake, write to [email protected].
5. Consequences
| Situation | What we may do |
|---|---|
| First minor or unclear violation | Block the request, show a warning, remove content |
| Repeated attempts, jailbreaking, harassment | Limit or turn off AI features, temporary suspension |
| Serious violation (for example sexual content involving minors, credible threats, terrorism, fraud) | Immediate permanent ban, content removal, preservation of evidence, and reporting to law enforcement or bodies such as NCMEC, as the law requires or allows |
We record safety events (the request category, time and account) to enforce this policy and may keep them after an account is closed to prevent return. Refunds of coins or fees are not given for accounts closed for violations, except where the law requires.
6. Reporting and appeals
- Report content or behaviour that breaks this policy to [email protected], with a link, the content title, or a screenshot. We review reports promptly and prioritise anything involving a child at risk.
- Immediate danger? Contact your local emergency number. To report child sexual abuse material, you can also contact your national hotline (in the US, the NCMEC CyberTipline at report.cybertip.org).
- Appeals. If your account was restricted or closed, email [email protected] within 30 days with your username and an explanation. A person who was not involved in the first decision will review it. Decisions about child-safety violations are normally final.
7. AI transparency and limits
- Content marked as AI-generated, and tutor replies, are written by AI models, not people, and may be wrong or inappropriate despite safeguards.
- Prompts, imports and replies may be processed by third-party AI providers and, where needed for safety, reviewed by us. See the Privacy Policy.
- We continuously improve our safeguards and may update this policy as new risks and laws emerge.
8. Contact
Safety reports: [email protected]. Everything else: [email protected].