Security

Anyone can join the Topo Network with a GPU, so no node is trusted by default. The network is designed around that: requests are encrypted and kept in memory, every operator puts stake at risk, a sample of outputs is re-run by independent attestors, and every response leaves a signed receipt.

• All traffic to the API, dashboard and website uses TLS. Requests stay encrypted between your application, the Topo Router and the serving node.
• Prompts and outputs are processed in memory and are not stored by default.
• Account data and usage metadata are encrypted at rest.
• On Max and Enterprise plans, region pinning and private node pools keep traffic in the region, or on the hardware, you choose.

• Every operator stakes TOPO before it serves traffic, and the value a node can serve at once is capped by its stake.
• A sample of requests is re-run by three independent attestors on different hardware: 20% for new nodes, 2% once established, and every request when you choose strict verification.
• A mismatch slashes the node's stake, and the request is served again on another node at no cost to you.
• Each receipt is signed with the node's ed25519 key and included in an on-chain batch root, so you can check both yourself.

The details are in the verification docs.

• Keys are shown once, when they are created, and can be scoped by model, monthly spend, region, IP range and expiry.
• A request outside a key's scope is rejected before any tokens are served.
• Rotation keeps the old key valid for a grace period you choose. Revoked keys stop working within 10 seconds in every region.
• For browser and mobile apps, mint short-lived tokens from your backend. They expire after 15 minutes and inherit the key's scopes.
• Max plans add team roles, and Enterprise adds SSO and SCIM provisioning.

The audit log records security-relevant actions, such as creating, rotating and revoking keys, changing scopes and spend limits, topping up the wallet and changing team members, with who made each change and when. Every request also has a receipt you can look up by its request ID.

• Production access is limited to the staff who need it, protected by multi-factor authentication and logged.
• The router, the node client and the verification spec are open source, so anyone can review how requests are served.
• Changes are code reviewed and tested before they are released.
• We monitor the network around the clock and follow a written incident response process. If an incident affects your data, we will notify you without undue delay.

If you believe you have found a security issue in Topo, please tell us. Email support@topolabs.ai with “Security” in the subject line, and include:

• a description of the issue and where it is;
• steps to reproduce it, or a proof of concept;
• the impact you believe it has;
• how to reach you, and whether you would like to be credited.

We will acknowledge your report, keep you updated while we investigate, and tell you when the issue is fixed. Please give us reasonable time to fix it before you share any details publicly.

In scope

topolabs.ai, docs.topolabs.ai, the dashboard, api.topolabs.ai, the node client, the SDKs and the settlement contracts.

Out of scope

denial-of-service or volumetric testing, social engineering and phishing, physical attacks, and issues in third-party services we do not control.

When you test, please:

• use only accounts and keys that you own, and prefer sandbox tp_test_ keys;
• never access, change or delete data that is not yours, and stop as soon as you reach someone else's data;
• avoid degrading the Service for other users, and do not attack nodes, attestors or wallets you do not control;
• keep the details of the issue confidential until it is fixed.

If you make a good-faith effort to follow this policy, we will consider your research authorized. We will not pursue legal action against you or report you to law enforcement for it. If a third party brings legal action against you over activity that followed this policy, we will make it known that your research was authorized.

This safe harbor covers our own systems and the claims we control. It cannot authorize testing of third-party systems. If you are not sure whether something is in scope, ask us first.

Security reports and questions: support@topolabs.ai, with “Security” in the subject line.