Security & data handling
A plain-English description of how Academe handles your research data, what we encrypt, who can read it, and where it lives.
No training on your data
Your data is not used to train AI models. Each model-backed request is scoped to the task: the necessary context is sent, a response comes back, and server-side retention is limited to what is needed to provide and operate the service.
We have agreements with model providers prohibiting training on your inputs or outputs.
Encryption
- In transitAll traffic between your browser and Academe uses TLS 1.2 or higher.
- At rest: databaseEvery row in our database is encrypted by the underlying storage provider.
- At rest: filesUploaded PDFs and attachments are encrypted in object storage.
Access control
Every database table is gated by row-level security. You only see your own projects and projects that have been explicitly shared with you. Database access is restricted to authenticated, authorized requests.
The only routes that bypass row-level security are background jobs running with a service role, and those are scoped to strict per-tenant boundaries.
Third-party AI providers
Academe routes writing, chat, review, embeddings, and transcription through contracted model providers. Those providers are prohibited from training on your data.
Provider calls happen over TLS and are scoped to the minimum context required for the task.
Data retention & deletion
Your projects, notes, and files persist until you delete them. Export your full account at any time as a ZIP archive from Settings → Privacy.
Where data lives
- ApplicationVercel (US regions), SOC 2 Type II audited.
- Database, auth, storageSupabase (US regions), SOC 2 Type II audited.
- LaTeX + real-time collabFly.io, SOC 2 Type II audited.
Reporting a security issue
Email hello@academe-ai.com with a description of the issue and reproduction steps. We respond within one business day.