← Back to the cybersecurity blog

Cybersecurity in schools and universities: what to protect and where to start

By Adrián González · CEO y socio fundador · Published: 21 November 2026 · Updated: 21 November 2026
Cybersecurity in schools and universities AI-generated image

On 2 July 2026 the University of Alicante's monitoring systems flagged suspicious activity that looked like the start of a ransomware attack. As a precaution the university shut down its UACloud intranet, staff remote access and its online services portal, pausing administrative deadlines for anything handled online. A year earlier, the education department of the Castilla y León region had disclosed a leak of pupil and family data from its attendance system.

Spain does not measure the problem by sector, but the UK does. Ninety-eight per cent of UK universities identified a breach or attack in the year covered by the government's Cyber Security Breaches Survey, published in April 2026, against 73% of secondary schools, 49% of primary schools and 43% of businesses overall.

This guide is for whoever is responsible for technology or compliance at a university or school, whether state, state-funded or independent. It uses Spain as a worked example within the EU framework, and covers which information is most sensitive, what the law requires and which measures to prioritise, with a dedicated section for schools that have no IT team.

Why do attackers target schools and universities?

In 2024 Microsoft ranked education as the third most targeted sector in its threat intelligence and counted more than 15,000 messages a day carrying malicious QR codes aimed at schools and universities. What draws attackers is less the money any one institution holds than the way these places work.

  • Thousands of identities that turn over every year: students arriving and leaving, temporary teaching staff, visiting researchers and alumni who keep their mailboxes.
  • Networks built to be open, with Wi-Fi for any device, services published to the whole community and personal laptops everywhere.
  • Very young or untrained users, who every day receive easily spoofed messages about grants, enrolment or learning platforms.
  • A long chain of education technology suppliers processing children's data on the institution's behalf.
  • At schools, tiny IT teams or none at all beyond the regional authority; at universities, decentralised IT, with departmental services and researchers running their own servers.

In the UK survey, phishing featured in 90% or more of the institutions that suffered an incident, and 29% of universities reported attacks at least weekly. The samples for universities and further education colleges are small, so those percentages are best treated as indicative.

Data protection in schools and universities: the most sensitive information

An educational institution gathers data of very different kinds, much of it about children. For schools below university level, Spain's Education Act allows such data to be collected where teaching requires it, including a pupil's family and social background and personal circumstances (twenty-third additional provision). The same provision bars using it for other purposes without express consent, and requires security measures and a duty of confidentiality from anyone who handles it.

What data an educational institution holds and why it matters
Type of informationWhere it usually livesWhy it is sensitive
Health, disability and special educational needsPastoral and counselling teams, school nurse, records, exam adjustmentsSpecial category data under the GDPR, and highly damaging if a child's record leaks
Psychological and behaviour reportsCounselling service, form tutors, teachers' mailboxesProfessional opinions about children that should never leave the school
Family circumstances: custody, restraining orders, safeguardingSchool office or registry, head's office, notes in the management systemA single access error can put a child at physical risk
Grants, bursaries and financial hardshipFinance and bursary offices, payment gatewaysFinancial data on vulnerable families, plus payment fraud
Marks, academic records and certificatesStudent information system, exam boards, degree registersTampering with marks and forged qualifications
Photos and videos of pupilsWebsite, social media, virtual classrooms, teachers' phonesPublishing them on the web or social media needs consent and is hard to undo
Biometrics and online exam invigilationRemote invigilation tools, access control systemsSpecial category data, with recent penalties against universities
Research data and intellectual propertyResearch groups, departmental servers, cloud storageThird-party clinical data, industry contracts and espionage

Hard2bit analysis based on Spanish education and data protection law and the regulator guidance cited in the text.

Credentials belong on that list too. One teacher's account reaches the marks and details of dozens of families, and an administrator in the school office or registry, or a principal investigator, can reach far more. In July 2026 Séneca, Andalusia's school management platform, suffered a possible unauthorised access. José Antonio Nieto, the regional minister for AI, Digital Development and Public Administration, put it down to an oversight by an authorised user on their personal computer and acknowledged that the scope was not yet known. The regional government asked every user to change their password.

Cybersecurity in universities: the ENS, NIS2 and research

Which legal framework applies

Spanish public universities are part of the institutional public sector under article 2 of Law 40/2015, so the National Security Framework (ENS), Royal Decree 311/2022, applies to them directly. In February 2022 the National Cryptologic Centre (CCN) published the CCN-STIC 881 guide on ENS adoption for universities alongside a sector compliance profile. The guide predates the current decree and should be read together with it.

Private universities fall outside the ENS, but article 2.3 of the decree extends it to the systems a private organisation uses to supply services or solutions to the public sector under contract. For everything else, ISO 27001 is the usual benchmark. Either way, the scope should take in the bodies that share a domain, network or identities with the university: foundations, technology transfer offices, spin-outs and science parks.

NIS2 does not list education as a sector. Annex II covers research organisations, but the definition in article 6(41) of the directive expressly excludes educational institutions. Article 2(5)(b) lets each member state bring them in, particularly where they carry out critical research.

Spain has yet to transpose the directive. On 8 July 2026 the European Commission referred Spain to the Court of Justice, and the dissolution of parliament ahead of the general election on 29 November stalls the parliamentary route. How universities are treated will depend on the eventual Spanish law; our NIS2 page tracks it.

Where campus risk concentrates

A university's attack surface spans the administrative network, the academic network and community Wi-Fi. On top of that come lab machines often left unpatched because they drive scientific instruments, and departmental servers a research group set up years ago.

Alicante shows what happens when a peripheral corner of that estate is hit. According to the university, the malware reached a few secondary servers, and the precautionary shutdown disrupted remote working and the online services portal. Its management says essential services were restored within about 24 hours and that subsequent checks found no data had leaked.

Three issues will be familiar to any university security lead. The first is eduroam: if a device does not validate the authentication server's certificate, a rogue access point can harvest the institutional username and password, which usually also unlock email. That is why eduroam should only be rolled out with profiles generated through eduroam CAT or geteduroam, ideally with Wi-Fi-specific credentials.

The second is the virtual learning environment. Moodle and similar platforms accumulate third-party plugins and integrations, each of them code that has to be kept up to date. The third is federated identity through RedIRIS's identity service (SIR), whose identity providers are not always Entra ID or Google: the second factor has to be enforced there and signalled to the applications that require it.

Research security

Research servers hold clinical data shared by hospitals, surveys containing personal details, results with commercial value and, in some fields, material of interest to foreign intelligence services. They are often missing from the central inventory and from security operations centre (SOC) monitoring, because the research group runs them.

The EU now addresses this risk head-on. The Council Recommendation of 23 May 2024 on enhancing research security asks universities to manage the risks of foreign interference and knowledge leakage, and Recommendation (EU) 2021/1700 gives guidance on internal compliance programmes for dual-use research. Both translate into very practical controls: who can reach the computing clusters and repositories, which accounts visiting researchers use, and what gets logged.

Remote invigilation, biometrics and the AI Act

Online exam invigilation is where security and data protection meet head-on. Spain's data protection authority, the AEPD, fined the Universitat Internacional Valenciana €650,000 (PS/00067/2024) for imposing facial recognition in online exams with no genuine alternative for students, and for processing more data than it needed. The authority also rejected the idea that current university legislation provided a sufficient legal basis for biometric processing.

In a separate case the AEPD set a penalty of €200,000 for the Universidad Europea de Valencia (PS/00447/2025) over a similar system: €50,000 for processing biometric data without valid consent and €150,000 for launching it without a prior impact assessment. The university paid €160,000 after the reduction for voluntary payment.

The EU AI Act brings rules of its own. Since 2 February 2025 it has banned systems that infer people's emotions in educational institutions, except for medical or safety reasons (article 5(1)(f)). Annex III treats as high-risk the systems used for admission, for assessing learning outcomes, for placing students at an appropriate level and for monitoring behaviour during tests. Regulation (EU) 2026/1744 has pushed those obligations back to 2 December 2027, which leaves time to review contracts but offers no shelter from the GDPR, the basis of both AEPD penalties.

Schools: children, families and platforms

Who answers for the data

At Spanish state schools, the data controller is the regional education department, which also appoints the data protection officer and falls within the ENS. At state-funded and independent schools, the controller is the organisation that owns the school, even when it uses the regional management platform, whose controller remains the department. A state-funding agreement is not, in principle, a public services contract, so it does not bring the ENS with it, and each party's role should be documented.

Spanish data protection law requires a data protection officer at every school teaching any of the levels set by education legislation, and at public and private universities (article 34.1.b of the LOPDGDD). Providers of non-formal tuition, such as language academies, fall outside that specific duty.

When consent is needed

Most of what a school does with pupils' data needs no consent. It rests on a task in the public interest or a legal obligation (article 6(1) GDPR), at independent schools also on the enrolment contract, and for health data on article 9(2)(g) read with the Education Act provision. Recording classroom activity for teaching and sharing it in a closed, authenticated environment does not need consent either, according to the AEPD.

Consent comes in for anything beyond that, such as publishing images on the website or social media, which article 92 of the LOPDGDD ties to protecting the best interests of the child. A child can consent for themselves from the age of 14, unless the law requires parental assistance for that act (article 7); below that age a parent or guardian consents. A bill that would have raised the threshold lapsed when parliament was dissolved in October 2026.

Digital learning platforms

In March 2026 the AEPD and the data protection authorities of Catalonia, the Basque Country and Andalusia published principles for procuring and using digital education platforms. They regard a data protection impact assessment as mandatory before launching any platform that processes children's data at scale in the cloud, and warn that a supplier's legitimate interest can hardly justify advertising, profiling or analytics on that data.

They also warn that unclear data flows to other countries are reason enough not to authorise a platform, and that a supplier using the data for its own purposes stops being a mere processor. The GDPR already obliges processors to report breaches to the controller (article 33(2)), but the contract should set a specific deadline so the school can meet its own. That clause is a basic element of supplier risk management in education.

What PowerSchool taught the sector

PowerSchool, one of North America's largest school management software suppliers, had pupil and teacher data from thousands of districts stolen at the end of 2024. According to CrowdStrike's findings as reported by SecurityWeek, the attacker gained access using compromised credentials for a maintenance account on its support portal. The perpetrator was sentenced in October 2025 to four years in prison for an extortion scheme in which, according to US prosecutors, he threatened to leak data on more than 60 million students and 10 million teachers.

The management platform supplier is a single point of failure: if its support access falls, every customer can fall with it. Ask how it is protected before you sign.

Baseline controls for a school with no IT team

A school with a part-time ICT coordinator is not going to run a SIEM. It can, however, close most of the doors with what is already in its Google Workspace for Education or Microsoft 365 admin console.

  • Two-step verification for all staff. In Microsoft 365 A1 the free security defaults switch it on; in Google Workspace it is enabled from the admin console.
  • Pupil organisational units flagged as under 18 in Google Workspace, which blocks unconfigured third-party apps by default, with only the approved catalogue allowed.
  • On Chromebooks, guest mode off, sign-in limited to domain accounts, forced re-enrolment after a wipe and user data erased at sign-out on shared trolleys.
  • In the school office, accounts without administrator rights, screens locked when staff step away, no passwords saved in shared browsers, and the regional platform never opened on unpatched personal devices.
  • A phone call to a number already on file before acting on any change of bank details or request for a pupil's information, even when it arrives by email or through the parent messaging app.
  • A clear rule on generative AI: staff do not enter identifiable pupil data into tools the school has not contracted, and approved tools are enabled per organisational unit and age group.

Schools seldom control their Wi-Fi infrastructure, which is usually run by the regional authority or the operator. What they can do is keep the guest network isolated, give pupils a separate key and keep the school office off the classroom network. Their internet exposure usually comes down to the school website and cameras or network drives published by mistake.

Cyberbullying and the spread of intimate images

In secondary schools, one of the most damaging incidents for a pupil is the spread of intimate images of them, real or manipulated with AI. Spain's child protection law, LOPIVI (Organic Law 8/2021) requires every school with under-18s to appoint a wellbeing and protection coordinator and to follow protocols against cyberbullying.

When such images circulate, the school should request their removal through the AEPD's priority channel, avoid forwarding or keeping copies, and work with that coordinator. INCIBE's free and confidential 017 cybersecurity helpline also supports teachers and families.

Universities and schools have different priorities

Risks and obligations by type of institution
AspectUniversitySchool
Security frameworkENS for public universities; ISO 27001, or ENS by contract, for private onesENS through the regional department for state schools; GDPR and LOPDGDD for all
Data controllerThe university itselfThe regional department for state schools; the owning organisation for state-funded and independent schools
IdentitiesTens of thousands, federated, including external researchers and alumniFewer accounts, many belonging to under-14s
Exposed surfaceVirtual learning environment, VPN, eduroam, online services portal, research serversRegional platform, virtual classroom, website and teachers' apps
Most sensitive dataResearch, health and exam biometricsChildren's health, counselling and family circumstances
Typical incidentRansomware, credential theft, intrusion into departmental serversCompromised teacher account, fraud against families, spread of images
Reference CSIRTCCN-CERT for public universities, INCIBE-CERT for private ones, plus RedIRIS's IRIS-CERTCCN-CERT via the regional department for state schools; INCIBE-CERT for state-funded and independent schools

Indicative only; each institution should adapt it to its own risk analysis.

Cybersecurity best practice for schools and universities: what to prioritise

In our view, these are the priority controls, in order of impact. Schools without an IT team will find a slimmed-down version in the previous section.

Identity: two-step verification and up-to-date accounts

The first priority is two-step verification for teaching and administrative staff, starting with the school office or registry, senior leadership, counselling and principal investigators. Students at university can reasonably be asked to use it too. Administrator accounts call for a phishing-resistant method such as FIDO2 keys or passkeys, enforced through a Conditional Access policy that excludes one emergency account so the tenant cannot be locked out.

Account lifecycle goes hand in hand with it. At a school, leavers' accounts should be closed or downgraded to minimal rights at the end of the year. A university cannot work to a fixed date, given resit sessions, dissertations and staff with several affiliations: closure should follow the live relationship recorded in admissions and HR, with a defined grace period. Alumni accounts and those of staff who have moved on are a frequent foothold, because their activity often goes unchecked.

The query below needs Entra ID P1 or P2, which come with the Microsoft 365 A3 and A5 licences but not with A1, and logs sent to Log Analytics or Microsoft Sentinel. It shows which accounts sign in without being required to use a second factor:

// Successful interactive sign-ins over the last 30 days, per user.
SigninLogs
| where TimeGenerated > ago(30d)
| where ResultType == "0"
| summarize SignIns = count(), NoMFARequired = countif(AuthenticationRequirement == "singleFactorAuthentication") by UserPrincipalName
// Filter on the staff domain if students and staff use different domains.
| where NoMFARequired > 0
| order by NoMFARequired desc

The query covers interactive sign-ins only; non-interactive ones sit in the AADNonInteractiveUserSignInLogs table. Without P1, the same check can be done in the Entra admin centre sign-in logs by filtering on the authentication requirement. In Google Workspace, the user log events report can be filtered for successful sign-ins without a second factor. Where a university runs its own identity provider federated through SIR, the equivalent review happens in that provider's logs and in RADIUS. Our Microsoft 365 security service includes this review for education environments.

Email and messaging

Beyond the provider's filters, publishing SPF, DKIM and DMARC under a policy that blocks spoofing of the institution's domain helps against phishing. Flagging external email clearly helps, as does teaching people to distrust QR codes that ask them to sign in. Training works best when it is short and repeated, and it supports the digital competence that article 83 of the LOPDGDD expects of teachers so they can teach pupils to use digital media safely.

Separate the networks

The student network and guest Wi-Fi should not be able to reach the school office, management servers or backups. At universities, research needs its own segment, with rules agreed with each group and a design built for large data transfers, because a firewall that becomes a bottleneck soon gets bypassed. Our guide to internal network segmentation explains how to go about it.

Backups that survive ransomware

Backups of the student information system, of email and of pupil records should include at least one immutable or offline copy, and restores should be tested before they are needed. For management systems, the 3-2-1-1-0 rule sums up the requirements: three copies, two media, one off site, one immutable or offline and zero errors in restore tests. Scientific data, which can run to petabytes, needs a data management plan agreed with each group instead.

Suppliers and classroom apps

Teachers pick up apps quickly, often free ones and with no contract. An approved app catalogue, reviewed by the data protection officer, curbs shadow IT without shutting down teachers' own initiatives. Every supplier handling children's data should have a processing agreement, an impact assessment where required, a breach reporting deadline, transparency about where data is stored and, in the public sector, a requirement to conform to the ENS.

Exposure to the internet and patching

VPNs, remote access portals, the virtual learning environment and departmental web servers are the other main exposure. An inventory of what the institution publishes to the internet, reviewed monthly, lets the team patch exposed systems first. An external check such as our attack surface scanner offers a first starting point.

Logging and the ability to respond

Without logs it is very hard to establish what an attacker took, and that largely determines whether families must be told. Entra ID keeps sign-in logs for between 7 and 30 days depending on the licence, so they have to be exported to be available months after an incident. At a minimum, an institution should keep sign-in records from its identity platform and email, plus access logs for the student information system.

Where there is no in-house team to watch them, a managed SOC can fill the gap. Spanish public universities can also draw on CCN-CERT's services and, where one exists, their regional CSIRT.

When an incident happens: the first 72 hours

The GDPR requires a personal data breach to be notified to the supervisory authority without undue delay and, where feasible, within 72 hours of the controller becoming aware of it, unless it is unlikely to pose a risk to individuals. In Spain that authority is the AEPD or, in Catalonia, the Basque Country and Andalusia, the regional authority where the processing belongs to the regional public sector, public universities included (article 57 of the LOPDGDD). For state schools, the regional department notifies. Where the risk is high, those affected must be told as well, and where children's data is involved families will very probably have to be informed.

  • Isolate what is affected without switching it off or wiping it, so that evidence survives.
  • Bring in the data protection officer within the first hour, since the clock runs from the moment the institution becomes aware.
  • Notify the reference CSIRT set by article 11 of Royal Decree-Law 12/2018: CCN-CERT for the public sector, which the ENS also requires to be told of incidents with significant impact, and INCIBE-CERT for private organisations. Universities on the RedIRIS network can also call on IRIS-CERT, which is often the first to spot and report a problem.
  • Messages for families and students, and for staff, drafted in advance, with one named person speaking for the institution.
  • If an account was the entry point, forced password resets and revocation of active sessions; with Séneca, the Andalusian government asked every user to change their password.

A plan written in advance shortens every one of these steps. Our guide to building an incident response plan works as a template, and Hard2bit's incident response service is available to institutions with a retainer in place.

Common mistakes

The most common is assuming the regional authority or the supplier takes care of everything. They answer for their own systems, but staff accounts, classroom devices and apps the school signs up to remain the school's responsibility.

The second is treating data protection and cybersecurity as separate fields. The AEPD's invigilation penalties show that choosing a technology tool is also a data protection decision, and every security breach ends up as a data protection problem.

The third is keeping data longer than necessary. Exam recordings, counselling files and health data from past years multiply the impact of any breach. Academic records have to be retained, but archived away from the systems used every day.

Where to start

This term's first task fits into an afternoon and has three steps. First, switch on two-step verification for every member of staff with access to pupil data. Next, check the sign-in logs for accounts still not required to use it and close those no longer in use. Finally, list the suppliers that process children's data and review the contracts behind them.

Two-step verification costs little and blocks much of the stolen-credential access seen in several of the incidents above, from Séneca to PowerSchool. Our education sector page explains how we support universities and schools along the way, from the ENS to incident response.

This article summarises legal obligations and good practice as of 21 November 2026 and is no substitute for legal advice or for each institution's own risk analysis. The incidents and penalties cited are based on information published by the institutions themselves, by authorities or by the media at that date, and mentioning them does not imply security flaws in third-party products they used. The KQL query and the settings mentioned are illustrative and should be tested in each organisation's environment before use.

Frequently asked questions

What cybersecurity measures should a school have as a minimum? ▾

Two-step verification for staff, closure of unused accounts, pupil accounts restricted by age, an approved app catalogue, school office computers without administrator rights, an isolated guest network, backups that cannot be deleted from the same machines, and a procedure for telling families and authorities about a breach.

Does Spain's National Security Framework apply to private universities or state-funded schools? ▾

Not directly. The ENS is aimed at the public sector and reaches a private organisation only in the systems it uses to serve a public body under contract (article 2.3 of Royal Decree 311/2022). In principle, a state-funding agreement is not that kind of contract. For its other systems, most institutions work to ISO 27001.

Do independent and state-funded schools need a data protection officer? ▾

In Spain, yes. Article 34.1.b of the LOPDGDD requires one at schools teaching any of the levels set by education legislation, whoever owns them. For state schools the officer is appointed by the regional department, which is the data controller.

Does NIS2 cover universities? ▾

Not directly for now. Education is not one of its sectors, and educational institutions are excluded from the definition of research organisation, although member states may bring them in under article 2(5)(b). Spain has not yet transposed the directive, so it will depend on the law it passes.

Can a university use facial recognition to invigilate online exams? ▾

Only with great care. Spain's data protection authority has penalised two cases: one for imposing it with no genuine alternative and collecting excessive data, the other for relying on invalid consent without an impact assessment. The AI Act also classes exam monitoring as high-risk, with obligations from December 2027.

Does a school need parental consent for everything? ▾

No. The public interest and the school's legal obligations cover most processing, and recording classroom activity for teaching in a closed environment does not need consent either. Consent is needed to publish images on the web or social media; in Spain pupils give it themselves from 14, and parents or guardians below that age.

What should we do if our learning platform supplier suffers a breach? ▾

The institution, or the regional department for state schools, remains the data controller. It should obtain the scope from the supplier, decide with its data protection officer whether the regulator must be notified without undue delay, and within 72 hours where feasible, and inform families if the risk is high.

Which Microsoft 365 licence do I need to see who signs in without a second factor? ▾

Exporting sign-in logs to Log Analytics or Sentinel and querying them with KQL requires Entra ID P1 or P2, included in Microsoft 365 A3 and A5. With A1 you can switch on the free security defaults and review sign-ins in the Entra admin centre, which keeps them for a limited time.

Share this article

Cite this article

If you write about this topic, you can cite it like this:

Adrián González (21 November 2026). Cybersecurity in schools and universities: what to protect and where to start. Hard2bit. https://hard2bit.com/en/blog/cybersecurity-education-universities-schools-sensitive-data/

<a href="https://hard2bit.com/en/blog/cybersecurity-education-universities-schools-sensitive-data/">Cybersecurity in schools and universities: what to protect and where to start</a> (Hard2bit)

Want a straight answer on scope, priorities and price?

Most organizations reach us mid-question: something needs fixing, nobody has scoped it, and finance wants a number. Thirty minutes with a technical consultant — not a salesperson — gets you a defined scope, priorities ranked by risk and a price range. With what comes out of that call, we turn it into a fixed proposal. We work with organizations in Spain, across the EU and in LATAM.

If your situation is different, tell us anyway — we also take one-off questions on cybersecurity and regulatory compliance.

Based in Spain · Working across the EU and LATAM · ENS High · ISO 27001 · We usually reply in under 24 business hours