AI Science Tutor
Safeguarding and Child Safety Policy
Version 1.0 | Last updated: 27 August 2026
This Policy explains the safeguarding and child-safety approach used for the AI Science Tutor operated by Mr Mark Scates, a sole trader trading as Online Science Tuition.
The AI Science Tutor is an educational service for school science. It may be used by children and teenagers, so the service is designed to place safety, privacy and the best interests of children at the centre of its operation.
This Policy applies only to the AI Science Tutor. It does not replace the separate safeguarding arrangements that may apply to private one-to-one tuition.
Important information for students and parents
• The AI Science Tutor is an educational tool. It is not an emergency service, counselling service, medical service or substitute for a trusted adult.
• The Tutor is not continuously monitored by a person. A student must not rely on the Tutor to report abuse, self-harm, immediate danger or another safeguarding concern.
• If a student is in immediate danger in the UK, they should call 999 or ask a trusted adult to do so. A young person in the UK can also contact Childline free on 0800 1111.
• The Tutor uses automated safety and moderation systems. These can refuse, limit or redirect unsafe or inappropriate requests.
• Serious safeguarding categories can create a limited safeguarding record and an administrator alert. The safeguarding record and alert do not contain the student’s full message.
• Ordinary Tutor conversations and uploaded files are not stored in the website’s safeguarding log.
• Parents and guardians should encourage children to avoid sharing unnecessary personal, medical or highly sensitive information with the Tutor.
1. Our safeguarding principles
1.1 We aim to operate the AI Science Tutor in a way that supports the welfare, privacy and best interests of children and young people.
1.2 We use data minimisation and high-privacy defaults. The Tutor is designed to request only the information reasonably needed to provide and secure the educational service.
1.3 Safety controls are intended to allow legitimate curriculum science while restricting content that could create an unreasonable risk of harm.
1.4 Safeguarding measures must not unnecessarily prevent age-appropriate science education. Topics such as puberty, human reproduction, health, disease, genetics and other legitimate curriculum material may be discussed in neutral educational language where appropriate to the selected course.
2. Who may use the AI Science Tutor
2.1 The AI Science Tutor may be used by school-age students as well as adults.
2.2 A paid subscription must be entered into by a person aged 18 or over. If the designated student is under 18, the student’s parent or legal guardian must enter into the paid subscription.
2.3 Each account is for one designated student. Account sharing is not permitted.
2.4 Complimentary access may be provided to selected students, but the same safety, safeguarding, account and acceptable-use requirements apply unless we expressly state otherwise.
3. Automated safety controls
3.1 User-written Tutor requests and supported image attachments are checked by automated safety systems before they are sent to the normal Tutor model.
3.2 The service uses more than one form of automated safety checking. This includes moderation for broad safety categories and a separate curriculum-safety assessment intended to distinguish legitimate school science from unsafe, abusive or clearly inappropriate use.
3.3 Candidate AI responses may also be checked before they are shown to the student.
3.4 Depending on the result of the safety checks, the Tutor may answer normally, provide only safe curriculum information, give a safety-focused response, refuse the request, or block the normal Tutor response.
3.5 Automated systems can make mistakes. A legitimate educational request may occasionally be refused or a concerning request may not be identified. Students and parents must therefore not treat the Tutor as a substitute for human safeguarding support.
4. Examples of content that may be restricted
4.1 The Tutor may restrict or refuse requests involving self-harm or suicide instructions, exploitation or abuse, sexual content involving minors, threatening or violent wrongdoing, weapons, harmful chemicals or pathogens, dangerous practical instructions, or other content that could create a serious safety risk.
4.2 The Tutor may also restrict personalised medical diagnosis or treatment requests. It can still explain appropriate school-science content about health, disease and the human body in general educational terms.
4.3 Attempts to bypass safety measures by using role-play, fictional scenarios, translation, encoding, hidden instructions, uploaded files or other methods are treated according to the underlying request rather than the way it is presented.
4.4 Safety restrictions are separate from copyright and service-security controls. A copyright refusal or ordinary fair-use restriction is not automatically treated as a child safeguarding event.
5. Safeguarding records and administrator alerts
5.1 Where a safety system identifies a safeguarding category that meets the service’s logging threshold, the website may create a safeguarding event record.
5.2 The safeguarding record is deliberately limited to metadata. It may include an event reference, date and time, severity level, category, Tutor activity, selected educational context, WordPress user ID and review status.
5.3 The safeguarding record does not store the student’s full message, the AI response or the uploaded file.
5.4 More serious safeguarding categories can generate an administrator alert email. The alert email contains limited metadata and an event reference; it does not contain the student’s full message, attachment content or IP address.
5.5 Lower-level sensitive or borderline events may be logged without an email alert. Serious events are designed to generate an alert, and the highest-risk prohibited content is designed to be blocked before the normal Tutor response as well as logged and alerted.
5.6 Safeguarding event metadata is retained for limited periods. Current default retention is 30 days for lower-level safeguarding metadata, 90 days for serious-concern metadata and 180 days for the highest-risk metadata. Test safeguarding records are retained for 7 days.
6. What an alert does – and does not – mean
6.1 An automated safeguarding alert is a safety signal, not a diagnosis, finding of fact or determination that abuse or harm has occurred.
6.2 The service is not staffed as a continuously monitored safeguarding helpline. An alert may not be seen immediately and must not be relied upon to obtain emergency help.
6.3 Because the service deliberately avoids storing the student’s full safeguarding message in its alert or safeguarding log, the information available to us following an alert may be limited.
6.4 Where reasonably appropriate and lawful, we may review the available safeguarding metadata, restrict or suspend an account, contact the adult account holder, seek professional advice, or make a disclosure where we are legally required or where an applicable lawful basis permits it in order to protect a person from serious harm.
6.5 We cannot promise that every safeguarding concern will be detected, that an alert will always be delivered, or that we will be able to identify or contact a person involved in a concern.
7. What students should do if they are worried or unsafe
7.1 The AI Science Tutor should not be used as the only way to tell someone about abuse, self-harm, suicidal feelings, exploitation, violence or immediate danger.
7.2 A student who feels unsafe should tell a trusted adult as soon as possible. This could be a parent or guardian, teacher, school safeguarding lead, doctor, police officer or another responsible adult they trust.
7.3 If there is an immediate danger or emergency in the United Kingdom, call 999 or ask a trusted adult to call 999.
7.4 Children and young people in the UK can contact Childline free on 0800 1111 or use the support options at childline.org.uk.
7.5 A student outside the UK should contact the emergency services or an appropriate child helpline in their own country if urgent help is needed.
8. Personal information and privacy
8.1 Students should not put passwords, payment-card details, government identification numbers or other unnecessary confidential information into the Tutor.
8.2 Students do not need to provide a personal medical history, diagnosis, test result or treatment information to ask ordinary curriculum-science questions.
8.3 Ordinary Tutor prompts and responses are not stored in a permanent WordPress conversation table. Safeguarding records contain limited metadata rather than the student’s full message.
8.4 Further information about personal information, retention, OpenAI processing, Stripe and individual data-protection rights is provided in the AI Science Tutor Privacy Notice.
9. Parents and guardians
9.1 Parents and guardians are encouraged to discuss appropriate use of AI with the student and to remind them that AI-generated information can be wrong.
9.2 Parents and guardians should encourage students to use the Tutor for legitimate educational purposes and to speak to a real person if they are worried, distressed or unsafe.
9.3 A parent or guardian who believes an account has been misused, compromised or used in a way that creates a safeguarding concern should contact us using the details in section 12.
9.4 A parent or guardian should not assume that the AI Tutor provides live supervision of a child’s online activity. The service does not provide continuous human monitoring of individual Tutor sessions.
10. Account restrictions and misuse
10.1 We may restrict, suspend or terminate access where reasonably necessary because of serious or repeated misuse, attempts to bypass safety or security systems, unlawful use, account compromise, fraud, or a material risk to users or the service.
10.2 Safety restrictions may be applied immediately where reasonably necessary. Where appropriate, we may ask the adult customer to explain or remedy an account problem.
10.3 The ability to use a legitimate curriculum topic does not permit users to turn that topic into harmful, exploitative, graphic, sexualised or dangerous requests.
11. Review and improvement
11.1 We may update safety prompts, moderation, technical controls, safeguarding categories, data-retention periods and this Policy as the service, technology, guidance or law develops.
11.2 Changes are intended to maintain or improve child safety while preserving legitimate science education.
11.3 We review the service with regard to the principles of the UK Information Commissioner’s Children’s Code, including the best interests of the child, high privacy by default, data minimisation, transparency and accessible ways to raise concerns.
12. Safeguarding contact
If you have a non-emergency safeguarding concern relating specifically to use of the AI Science Tutor, contact:
Mr Mark Scates trading as Online Science Tuition
Sole trader
155 Ford Lane, Rainham, Essex, RM13 7BA, United Kingdom
Email: markscates486@yahoo.com
Telephone: 07722 510681
This contact is not an emergency service and is not continuously monitored. In an emergency, use the appropriate emergency service instead.
AI-Science-Tutor-Safeguarding-and-Child-Safety-Policy-v1.0 (1)