Healthtech & health data

Your AI agent writes into patient records
with no defined perimeter.

Clinical scribes, appointment booking, triage: healthcare agents read external content and write into sensitive systems. We test what an agent can do that it shouldn't, on your test environment, never in production.

Applicable regulations

HDS 2.0

May 16, 2026

Health Data Hosting: certification required by May 16, 2026 for any entity hosting health data on behalf of third parties.

GDPR: sensitive data

In force

Health data is sensitive data (Art. 9). Unauthorized exposure triggers a mandatory CNIL notification within 72 hours.

CNIL: priority sector

Ongoing controls

Healthcare is a CNIL priority enforcement sector. Cegedim Santé: 15.8M patient records exposed, €800K fine in 2024.

Common healthtech vulnerabilities

  • APIs returning patient data without proper authentication
  • Missing RLS policies on databases (Supabase, Firebase)
  • Unauthenticated webhooks that let anyone create admin accounts
  • Medical documents stored in public buckets
  • Sensitive data moving in the clear, at rest and in transit

On a medical training platform, we found unauthenticated admin account creation, giving full access to every user's data in under two minutes.

FAQ

Want to know what your AI agent can do?

Tell us about your agent, its tools, and client context. We will come back with the right review scope.

Discuss your audit