Monitor, control and safeguard students who use AI tools
Students are increasingly using and experimenting with AI in class and it’s becoming more challenging than ever to manage and safeguard them online.
But… with classroom.cloud’s safeguarding tool, staff can monitor students’ use of AI platforms/chatbots such as ChatGPT, Copilot, Gemini, Claude, Perplexity, and others – and be alerted if they’re using them in ways that raise concerns.
Requirements and guidance for schools using AI
Guidance has come into effect across the UK and EU to ensure schools are managing AI appropriately.
Ofsted is actively observing how schools adopt AI, expecting clarity in roles and responsibilities, and evidence of safeguarding around AI deployment.
The EU AI Act (Regulation 2024/1689) came into force in August 2024, creating a regulatory precedent.
The UK government has published updated “Generative AI: product safety standards” guidance for education settings. (January 2026)
UK Safer Internet Centre filtering guidance (July 2025) reinforces existing monitoring legislation and sets out guidance on real-time AI content analysis and pre-approval risk assessments.
KSCIE 2026 is released for the 2026-27 school year.
Para 163: “contact: being subjected to harmful online interaction with other users or generative AI applications that simulate this; for example: peer to peer pressure, commercial advertising and adults posing as children or young adults with the intention to groom or exploit them for sexual, criminal, financial or other purposes, conduct: online behaviour that increases the likelihood of, or causes, harm; for example, making, sending and receiving explicit images, including those generated using AI and online bullying…”
Para 175: The department has published Generative AI: product safety expectations to support schools to use generative artificial intelligence safely and explains how filtering and monitoring requirements apply to the use of generative AI in education.
Restrict usage
You can now add URLs for AI platforms/chatbots to your ‘restricted’ website lists, as well as restrict specific installed programs to only allow approved apps during lessons.
Schools can also monitor across all applications including Office 365 and Teams (channels and chats), and the CoPilot integration also allows schools to capture Copilot triggers when a student is signed into Copilot using the school’s account – as well as monitor students’ Copilot chats.
If AI sites are restricted only during specific subjects/tests but are allowed for research activities later, then see below how to safeguard students.
Monitoring
Schools can specify when you want the students to be monitored, for example, only during school hours or at all times when the student is using their school device or on the school network.
Keyword and phrase monitoring
classroom.cloud comes with preloaded 20,000+ safeguarding keyword packs that cover a wide range of risk areas, including self-harm, bullying, radicalization, and sexual exploitation. This is particularly useful when students use AI platforms to discuss sensitive or dangerous topics. The keyword list is constantly updated with the help from existing customer feedback, the IWF, UKSIC, SWGFL and consultants to ensure new terms and trends are captured.
When a keyword is triggered, classroom.cloud’s contextual Risk Index assesses the severity by looking at the nature of the phrase, the frequency of related behavior, and any previous incidents involving the same student. This enables staff to prioritize urgent cases rather than being overwhelmed by every mention.
Schools can also filter to only view those captured in an AI tool or website. This helps safeguarding staff understand the context of the trigger and identify activity involving AI tools. In the Activity tool, schools can also apply an overnight filter to check for overnight activity on student school devices. This can help highlight potential safeguarding concerns early on.
“The Trust has 5000+ devices running either Chrome OS and Windows…We find the ease of filtering through safeguarding concerns and the ability to visualise themes using the word cloud really impressive; it saves a lot of time.”
Evidencing the incident
Evidence is captured automatically depending on the severity. Low-severity incidents generate a text log with a timestamp and device information. High-severity incidents trigger a screenshot, while Urgent incidents take a screenshot and recording for complete context.
All evidence is stored securely, with details such as the student’s name, device, date, time, triggered keyword, and risk score, allowing staff to review incidents in detail.
The Image Analysis feature in the enhanced safeguarding toolkit helps to monitor and detect inappropriate images displayed on students’ screens from websites, apps and media – including generative Ai where images may be created.
Alerts
Real-time alerts are sent to designated safeguarding staff as soon as a concern is detected. This means they can immediately review the evidence and take swift action if needed, such as messaging the student or intervening in person. This system works whether the student is in the classroom, elsewhere in the school, or learning remotely.
“Every student in our school has a Chromebook; we currently have around 27 in use. The main way we use it is to monitor the pupils while teaching staff aren’t around, such as during break and lunchtimes when they…use YouTube or talk to different classrooms across the school. This enables us to monitor remotely, giving them a bit of freedom and independence while ensuring we keep track of what they are doing.”
Trending issues
classroom.cloud also helps schools spot patterns and trends through features like the word cloud view, which shows frequently triggered keywords across the school. This can highlight a rise in attempts to access AI tools or a surge in discussions about certain concerning topics. Such insights can feed directly into PSHE lessons, digital citizenship training, and policy updates.
Frequently asked questions
How is NetSupport evolving to address emerging AI-related risks?
Schools can monitor activity across leading AI platforms including ChatGPT, Microsoft Copilot, Gemini, Claude and Perplexity, with real-time safeguarding alerts when concerning behaviour is detected. The platform captures the AI application involved, monitors Microsoft Copilot conversations when students are signed in with their school account, and allows safeguarding staff to filter events specifically from AI tools. AI interactions are analysed using the Risk Index to provide contextual alerts, while evidence such as text logs, screenshots or screen recordings is automatically captured based on severity. Schools can also restrict access to AI websites during specific lessons or assessments while allowing approved educational use, helping them balance innovation with safe and responsible AI adoption.
What measures are in place to identify and categorise new AI tools and services quickly?
NetSupport are consistently reviewing emerging technology alongside feedback provided by existing customer base to identify what additional coverage or depth in monitoring we can provide – for example we currently detect 30 different AI websites that are known to ensure the best coverage.
How frequently are filtering categories, threat intelligence feeds and block lists updated?
Our 20,000+ Keyword list is consistently reviewed internally based on feedback/data from existing customers and with the support of IWF, UKSIC and consultants. The changes to the phrase library is updated automatically for end users with every new release of classroom.cloud (generally between 4-10 week frequency dependant on release).
What intelligence sources and safeguarding partnerships inform these updates?
Existing customer feedback, IWF, UKSIC, SWGFL, internal safeguarding professionals and MAT governors.
How does NetSupport identify and respond to AI tools operating within trusted domains or dynamically generated content?
Safeguarding triggers apply across all monitored content, regardless of whether it originates from an AI tool or another source. When a trigger is detected, classroom.cloud provides clear contextual information, including what was accessed, when it occurred and where it happened. Dedicated AI source identification adds further context by highlighting activity originating from supported AI platforms.
For schools using the Enhanced Safeguarding module, Image Analysis extends protection to AI-generated images, as well as images received through email attachments or transferred via USB devices. This helps identify potentially concerning visual content that text-based monitoring alone may miss.
Monitoring is available both on and off site, ensuring consistent safeguarding wherever students are learning. It also extends to Microsoft Copilot and Microsoft Teams through tenancy-based monitoring, providing visibility across Microsoft 365 environments.
Are there specific AI categories, policies or controls available to customers?
There are currently no AI-specific phrase match categories within classroom.cloud. However, activity from supported AI platforms is clearly identified, allowing safeguarding staff to distinguish AI interactions from other online activity. Schools can then apply the platform’s existing safeguarding controls, including real-time alerts, contextual Risk Index analysis, evidence capture and access controls, to monitor and manage AI use effectively.
In NetSupport view, does it currently align with the latest DfE filtering and monitoring standards and generative AI guidance?
Details for our KCSIE guidance/alignment can be found here: https://classroom.cloud/kcsie/
Are there any roadmap items, enhancements or best practice recommendations that schools should be considering over the next 12–24 months?
Schools should first ensure they are making full use of classroom.cloud’s latest AI monitoring capabilities, including AI source identification, Microsoft Copilot monitoring, contextual safeguarding alerts and Image Analysis where available. These features provide greater visibility into students’ use of generative AI and strengthen safeguarding coverage both on and off site.
While NetSupport does not publish its product roadmap, we work closely with schools and trusts to review their current classroom.cloud deployment, share best practice, deliver training and discuss future developments. We also actively welcome customer feedback, helping shape future enhancements based on the evolving needs of schools.
For trusts looking to strengthen their safeguarding provision further, Assisted Monitoring is worth considering. Delivered in partnership with SWGfL, it provides additional expert support by reviewing safeguarding alerts, helping reduce the workload on DSLs and safeguarding teams while ensuring potential concerns are identified and escalated promptly. This can free up valuable time and resources to focus on wider safeguarding priorities.
By providing both proactive monitoring and detailed incident evidence, classroom.cloud enables schools to demonstrate that they actively identify and address online risks, even when these occur through AI platforms.
classroom.cloud
classroom.cloud
