Skip to content
Layer

Security at Layer

Your work stays yours. Secure by default.

Layer is SOC 2 Type II certified. Customer data is never used to train foundation models, and every studio controls who gets in, what they can use and what they did.

Type II certified
SOC 2
Customer data residency
EU
Training on your data
Zero
To restore a deleted workspace
14 days

Data privacy

What happens to your data, and what never does

The short version of our Terms of Service and Privacy Policy, for the reviewer who needs it first.

  • No training on your content

    Customer data is never used to train foundation models. Layer does not train models on your content or inputs, and does not let any model provider train on them.

  • Data residency

    Customer data is stored in the European Union: your workspaces, the files you upload and generate, and your account.

  • Workspace isolation

    Each workspace's data is kept to that workspace, with no mixing of data between customers.

  • You own what you make

    Every asset you generate is yours, and its metadata is exportable at any time.

  • Retention and deletion

    Enterprise workspaces set their own retention: rules that delete files past an age you choose, every night. A deleted workspace or account can be restored for 14 days, then it is permanently deleted.

  • Subprocessors and DPA

    The subprocessor list is on trust.layer.ai, and a data processing addendum is available for studios that need one.

Security measures

Every control, in one place

What a security questionnaire asks, answered.

Compliance

4 controls

  • SOC 2 Type II

    Independently audited; the report is available under NDA

  • Trust center

    Subprocessors, security reports and compliance documents on trust.layer.ai

  • Data processing addendum

    Layer acts as processor for the personal data your studio controls

  • MSA review

    Master Service Agreement review as part of enterprise procurement

Identity & access

4 controls

  • Single sign-on

    Google, Microsoft, Okta and Azure AD out of the box

  • SCIM provisioning

    Users provisioned and de-provisioned from your identity provider

  • Role-based access control

    Roles and groups decide who can do what in each workspace

  • Approval controls

    Admins approve models, projects and budgets before they go live

Audit & monitoring

5 controls

  • Full audit trail

    Every creation, edit and run logged with a timestamp and the person behind it

  • Exportable event logs

    Download the complete log for compliance and regulatory review

  • Activity monitoring

    A live feed of the whole team's history

  • IP traceability

    Every asset traced back to the run, the model and the person that made it

  • Content Credentials (C2PA)

    Generated images, video and audio carry a signed C2PA manifest that marks them as AI-generated and made with Layer

Data & model governance

6 controls

  • No training on your data

    Your content never trains foundation models

  • Encryption at rest and in transit

    Data encrypted in storage, and over TLS on the wire with HSTS on layer.ai

  • EU data residency

    Workspaces, files and accounts stored in the European Union

  • Workspace isolation

    No cross-tenant data mixing; child workspaces keep their own assets and members

  • Custom data retention

    Enterprise: per-workspace rules delete files past a set age, and a parent can require them in every child

  • Admin-controlled model access

    Choose which AI models and modalities each workspace can use

Compliance

SOC 2 Type II certified

Layer is independently audited against the SOC 2 Type II trust principles. Enterprise customers can get the full audit report under NDA as part of procurement.

The subprocessor list, security reports and compliance documents are published on trust.layer.ai.

Responsible disclosure

Found a vulnerability? Tell us first.

Email security@layer.ai with a detailed report of the vulnerability and the steps to reproduce it, and please hold off on disclosing it publicly until we have responded and fixed it.

Frequently asked questions

Is Layer SOC 2 Type II certified?+

Yes. Layer is independently audited and SOC 2 Type II certified. The full audit report is available to enterprise customers under NDA as part of procurement; other compliance documents are on trust.layer.ai.

Does Layer train AI models on my content?+

No. Customer data is never used to train foundation models. Layer does not train models on your content or inputs, and does not let any model provider train on them.

Is our data encrypted?+

Yes. Customer data is encrypted at rest and in transit: everything Layer stores is encrypted at rest, and every connection to Layer uses TLS, with HSTS on layer.ai.

Where is my data stored?+

Customer data is stored in the European Union. When you run a model from a third-party provider, the prompt and files for that run are sent to that provider to produce the result.

Who are Layer's subprocessors?+

The current subprocessor list is published on our trust center at trust.layer.ai.

Do Layer's outputs carry provenance?+

Yes. Generated images, video and audio carry a signed C2PA manifest that marks them as AI-generated and made with Layer. The manifest is written into the file before it is published, in every format that has a C2PA container, and it records the workspace that made the asset but no personal data.

Can we control how long Layer keeps our files?+

Yes, on the Enterprise plan. Workspace admins set rules that delete generations and uploads past an age they choose, every night, while keeping files that are labeled, in a folder, liked or pinned. A parent workspace can require its rules in every workspace under it. Deleted files go to Trash, where an admin can restore them, and every run is recorded in the audit log.

Can we sign a data processing addendum?+

Yes. When Layer processes personal data on your studio's behalf, a data processing addendum governs it. Contact our sales team to put one in place.

How do I report a security vulnerability?+

Email security@layer.ai with a detailed report and steps to reproduce, and hold off on public disclosure until we have responded and fixed it.

Still have questions? Talk to our team

Security review

Everything your security team needs to say yes.

Reports, policies and subprocessors on trust.layer.ai, and a team ready for your questionnaire.