Safeguarding

Built for universities. Safe by default.

Augnovo classifies every question before it answers. Restricted topics are blocked, and the student sees a supportive response pointing to real help.

Online Safety Act 2023 Prevent duty Equality Act 2010 UK GDPR

How a question is handled

Every message runs through the same three steps before a single word is generated. Nothing is answered first and then checked afterwards.

01 Classify

The question is matched against each restricted category, using trigger terms and a context classifier.

02 Block

If it matches a category that is switched on, the assistant does not answer the question at all.

03 Respond

The student gets that category’s approved wording, including the support routes you have set.

Restricted categories

Each category has its own approved wording, shown to the student exactly as written below. Self-harm and Prevent are always on and cannot be switched off.

Explicit / NSFW

Strict
Detects: Around 30 trigger terms covering adult, explicit and sexual content, plus a classifier for context. Basis: University Acceptable Use Policy · Online Safety Act 2023
What the student sees I can’t provide explicit or adult content. This is restricted under the University’s Acceptable Use Policy and the UK Online Safety Act 2023.

If you need support with wellbeing and relationships, please contact Student Support: your Student Support link
  • Block the answer
  • Signpost Student Support

Reference: Content Category – NSFW

Self-harm & suicide

Always on
Detects: Around 15 trigger phrases covering intent, methods and self-harm, plus a classifier for indirect disclosure. Basis: Safeguarding duty of care
What the student sees It sounds like you’re going through a really tough time, and it’s important you get support right away. You are not alone.

Please talk to someone now:
UK: Samaritans – call 116 123 (24/7)
University Wellbeing Team: your wellbeing team number
In an emergency: call 999

Your safety matters, and support is available right now.
  • Block the answer
  • Signpost national and university support routes
  • Never return methods, under any phrasing

Reference: Safeguarding Response

Prevent – extremism & terrorism

Always on
Detects: Around 15 trigger phrases covering recruitment, propaganda and attack planning, plus a classifier for intent. Basis: Counter-Terrorism and Security Act 2015 (Prevent duty)
What the student sees I can’t provide information or support for extremist or terrorism-related content. This is restricted under the Counter-Terrorism and Security Act 2015 (Prevent Duty).

For academic study on this topic, please access balanced and approved resources via the University Library: your library link
  • Block the answer
  • Point to approved academic resources
  • Do not debate or negotiate the refusal

Reference: Prevent Duty

Harassment, hate & bullying

Standard
Detects: Classifier-led: slurs and targeted abuse across protected characteristics, threats of harm, doxxing and sustained harassment. Basis: University Code of Conduct · Equality Act 2010
What the student sees I can’t generate content that is harassing, threatening, or hateful towards a person or protected group. This is not permitted under the University Code of Conduct and the Equality Act 2010.

If you are experiencing harassment, please contact: your Report & Support link
  • Block the answer

Reference: Harassment Prevention

Drugs & weapons

Standard
Detects: Trigger terms covering production and sourcing of controlled substances and weapons. Basis: University policy · UK law
What the student sees I can’t provide instructions or help sourcing drugs or weapons. This is restricted under University policy and UK law.

If you need support with substance use, please contact Student Wellbeing: your Student Wellbeing link
  • Block the answer

Reference: Prohibited Goods

Academic integrity

Configurable
Detects: Requests to produce assessed work: assignments, dissertations and exam answers. Basis: University Academic Integrity Policy
What the student sees I can help you with research, explaining concepts, and improving your structure, but I can’t write your assignment, dissertation or exam answer for you, as this would breach the University’s Academic Integrity Policy.

Try asking me: “Explain this topic” or “Give me feedback on my draft”. your academic skills link
  • Block the answer
  • Redirect to permitted study support

Reference: Academic Integrity Policy

Personal data

Standard
Detects: Requests for personal details of students or staff, and for surveillance or CCTV material. Basis: UK GDPR · Data Protection Act 2018
What the student sees I can’t share personal information about students or staff. This is protected under UK GDPR and the Data Protection Act 2018.

Please contact the Data Protection Officer via your DPO contact page for data requests.
  • Block the answer

Reference: GDPR Protection

Wording shown like this is set per institution during onboarding, so students reach your own services. National routes are fixed and never configurable: Samaritans on 116 123 and emergency services on 999.

Configured with you, not by you

Filters are set up with your safeguarding and compliance teams when your assistant is built, so the wording, the support routes and the categories match your own policies.

Always on

Self-harm and Prevent filtering are built in and stay on. Neither is treated as a preference, and neither can be turned off for an institution.

Set at onboarding

The remaining categories, including whether the assistant helps with academic writing, are agreed with you before the assistant goes live.

Your wording, your services

Each response points to your own wellbeing team, Report & Support route and Data Protection Officer, alongside the national lines.

The duties this maps to

Wording is written to sit alongside the obligations your institution already reports against.

Online Safety Act 2023 Duties around harmful content for services accessible to students.
Counter-Terrorism and Security Act 2015 Restricted wording for extremism and terrorism-related requests.
Equality Act 2010 Protection from harassment and discrimination across protected characteristics.
UK GDPR & Data Protection Act 2018 Lawful handling of student and staff personal data, including logging.

To be clear about what happens behind the scenes: blocked questions are not reported to anyone, and no separate safeguarding record is created. Conversations are kept in your own conversation history in the usual way, so a blocked question stays there like any other message, visible to the staff who already have access. Augnovo does not use your data to train AI models, and one institution’s data never touches another’s.

Bringing Augnovo to your campus?

We’ll walk your safeguarding, IT and compliance teams through the filters and the wording — and tailor the responses to your own support services.

Talk to our team