Skip to main content

Compliance

Built for European regulatory requirements

Privacy-by-design, human oversight, and careful alignment with EU frameworks, not marketing slogans.

Posture

Trust is a product requirement

NeuraXplore designs PALS for classrooms and care settings where biosignals, adaptive AI, and educator authority intersect. These principles guide every product decision.

Privacy-by-design

Handling learner and patient-adjacent data requires the highest standard of protection. Webcam streams are processed locally in device memory where edge processing is used. Video is not stored or transmitted to the cloud. Only anonymised numerical proxies for cognitive load feed the adaptation engine.

EU AI Act readiness

Architecture is shaped around EU AI Act expectations for educational technology. We exclude emotion recognition and biometric identification. Signals stay tied to task-related interaction patterns and physiological proxies for information-processing capacity.

Human-in-the-loop

AI does not make autonomous educational or clinical decisions. It acts as an intelligent assistant so teachers, mentors, and therapists retain pedagogical and professional authority.

Data lifecycle

Minimal data, clear boundaries

Procurement teams and DPOs need to know what leaves the device, what stays pseudonymised, and who remains accountable.

01

Capture at the edge

Eye-tracking and pupillometry run on-device during a session. Raw video remains in working memory for inference, not archival storage.

02

Derive task-level scores

The adaptation layer receives anonymised numerical scores about attention and cognitive load, not identity-linked video or audio.

03

Pseudonymise and encrypt in transit

Where cloud services are required, data is pseudonymised and encrypted in line with GDPR expectations for educational deployments.

04

Keep humans accountable

Insights surface to educators and authorised staff. Outputs are framed for pedagogy and mentoring, not clinical labelling.

AI governance

Guardrails for adaptive systems

Adaptive PALS products share a common governance model: transparent signals, professional oversight, and explicit exclusions.

What we design for

Strictly non-medical biosignals

Biosignals adapt learning environments and supportive profiles. They are not used to diagnose, label, or replace professional assessment.

No high-risk emotion recognition

We do not infer emotion categories or identity from biometrics. Measurement stays tied to task performance and cognitive load proxies.

Faculty and mentor authority

AI recommends adaptations and content paths. Professionals decide what reaches learners and how insights are acted on.

Supportive, non-stigmatising outputs

Profiles and dashboards are written for classrooms and mentoring, not for surveillance or punitive scoring.

Transparency for institutions

We document data flows, retention posture, and oversight expectations so schools and partners can run their own DPIAs.

Evidence-led iteration

Research partnerships inform how signals are validated before they influence learner-facing experiences.

Regulatory mapping

How frameworks inform the architecture

Certification marks are not claimed until counsel confirms defensible wording. The table below maps how we align practices, not slogans, to European expectations.

GDPR

Data protection by design

Pseudonymisation, purpose limitation, encryption, and data-minimisation defaults for learner data processed in PALS deployments.

  • Pseudonymised learner identifiers
  • Purpose limitation for session data
  • Encryption for data in transit
  • Data-minimisation defaults
EU AI Act

Risk-aware educational AI

Human oversight, logging, transparency, and exclusion of prohibited practices such as emotion recognition in classroom contexts.

  • Human-in-the-loop decision paths
  • Activity logging for adaptations
  • Transparency documentation
  • No emotion or identity biometrics
AVG

Dutch and EU privacy expectations

Alignment with AVG interpretations for schools and Dutch institutional buyers alongside broader EU privacy law.

  • School and DPO-facing documentation
  • Dutch institutional buyer alignment
  • Parent and educator transparency
  • Retention posture for deployments

Practice alignment matrix

On-device video processing (no cloud storage)

GDPRAVG

Pseudonymised learner identifiers

GDPRAVG

Encryption for data in transit

GDPRAVG

Human oversight of AI outputs

EU AI Act

No emotion or identity biometrics

EU AI Act

Activity logging for adaptive decisions

EU AI Act

Institution-facing data flow documentation

GDPREU AI ActAVG

Purpose limitation for learner data

GDPRAVG

Checks indicate architectural or operational alignment documented for institutional review, not legal certification or compliance claims.

Across PALS

Product-specific safeguards

Each PALS product inherits the same compliance posture while emphasising different oversight surfaces.

NeuroPALS

Real-time biosignal adaptation with edge-first processing and classroom-safe profiling before labels.

NeuroPALS trust section

PALS Cortex

Faculty-approved courseware generation with human-in-the-loop publishing and EU AI Act-aware content workflows.

PALS Cortex trust section

PALS Pathway

Student-guided career exploration with mentor oversight, pseudonymised progress, and non-clinical framing.

PALS Pathway trust section

MindForge

Immersive learning environments with the same privacy-first defaults and professional authority model.

MindForge trust section

Work with us

Need a compliance conversation?

Share your procurement checklist, DPA requirements, or institutional review questions. We will walk through data flows, oversight, and documentation posture.