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.
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.
Derive task-level scores
The adaptation layer receives anonymised numerical scores about attention and cognitive load, not identity-linked video or audio.
Pseudonymise and encrypt in transit
Where cloud services are required, data is pseudonymised and encrypted in line with GDPR expectations for educational deployments.
Keep humans accountable
Insights surface to educators and authorised staff. Outputs are framed for pedagogy and mentoring, not clinical labelling.
Raw video · working memory only
Anonymised numerical output
GDPR-compliant transit
Educator view · no clinical labels
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.
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
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
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
| Practice | GDPR | EU AI Act | AVG |
|---|---|---|---|
| On-device video processing (no cloud storage) | GDPR | AVG | |
| Pseudonymised learner identifiers | GDPR | AVG | |
| Encryption for data in transit | GDPR | AVG | |
| 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 | GDPR | EU AI Act | AVG |
| Purpose limitation for learner data | GDPR | AVG |
On-device video processing (no cloud storage)
Pseudonymised learner identifiers
Encryption for data in transit
Human oversight of AI outputs
No emotion or identity biometrics
Activity logging for adaptive decisions
Institution-facing data flow documentation
Purpose limitation for learner data
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 sectionPALS Cortex
Faculty-approved courseware generation with human-in-the-loop publishing and EU AI Act-aware content workflows.
PALS Cortex trust sectionPALS Pathway
Student-guided career exploration with mentor oversight, pseudonymised progress, and non-clinical framing.
PALS Pathway trust sectionMindForge
Immersive learning environments with the same privacy-first defaults and professional authority model.
MindForge trust sectionWork 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.