This page sets out how we protect the data you trust us with on the AURÄI platform: what we promise, what happens to your data and for how long, how our connectors reach your systems, the security controls in place, and who processes data for us.
It describes what is true today. Where something is still being built, we say so.
1. Our commitments
- We do not train AI models on your data, and the AI model providers we use have committed to zero data retention: they don’t store what we send them or train on it, beyond brief in-memory caching while they answer.
- Evidence files you upload are deleted automatically, about 12 hours after upload by default. Settings shows your workspace’s retention period.
- Your workspace is kept apart from every other workspace: its data can only be read inside it.
- When Shadow AI opens in January 2027, its connectors will only ever read: never changing anything in your systems, and with access you can revoke at any time.
- We do not sell your information, and we do not use it or your workspace content to advertise to anyone.
2. Zero retention by AI model providers
When you run an agent, only what that run needs leaves your workspace: your instructions and the evidence you chose for it. It is processed by an AI model provider that has committed to zero data retention: the provider uses the request to produce the answer, doesn’t store it beyond brief in-memory caching while it answers, and does not use it to train models. We never fall back to a provider without that commitment.
We keep the answer with the run so you can read it, and we keep your own data only as long as the table below says.
| Data | How long we keep it |
|---|---|
| Evidence files you upload | Deleted automatically, about 12 hours after upload by default. Settings shows your workspace’s retention period. |
| Text of the evidence an agent run cites | 24 hours, so you can check the run’s citations, then removed |
| Agent runs and reports | Until you or your organization delete them, or the account is closed |
| Account, learning and billing records | While the account is open. After it closes we delete or anonymize them, except what we must keep for legal obligations such as tax and payment records, to resolve disputes, and to keep confirming certificates we issued. |
| Counters that limit sign-in and other attempts | Up to 24 hours, stored as keyed hashes rather than the email or network address itself |
| Support conversations, analytics and error reports | For the period set in each tool |
3. Connector access (Shadow AI, opening January 2027)
Shadow AI connects to your systems to find the AI tools, agents and API keys that can reach your organization’s data. It opens in January 2027; this section describes how it will work, and we will update it when it opens.
- Read-only. AURÄI only reads: it never changes anything in your systems, whatever a permission would allow. Fixes are steps you take in your own consoles.
- Settings and records, not your content. It does not read emails, chats, documents, prompts or secret values. The only files it reads are the dependency, AI tool configuration and workflow files in the GitHub repositories you choose.
- Deleted within an hour of the session ending. A scan session lasts at most 24 hours. When it ends, what it collected and every finding worked out from it are deleted within the hour. Your AI register is saved in a signed file that you keep, not us.
- What we keep. Your connection settings, any API keys you choose to save (encrypted; you can enter them each session instead), and a record of which systems each scan called and when, never what they returned. You will be able to see and export that record for each scan.
Every permission each connector asks for is public, with what it is for, what it never asks for and how to revoke it:
- Microsoft 365 and Entra ID20 read-only Microsoft Graph permissions, granted by an administrator.See every permission
- AWSA read-only role you create from our template, limited to listing and describing.See every permission
- GitHubA GitHub App with eight read-only permissions, on the repositories you choose.See every permission
- OpenAI API platformAn admin key you create, limited to reading, which we only read with.See every permission
- Anthropic ConsoleAn admin key you create, which we only ever read with.See every permission
- DNSPublic records only. Nothing to grant and nothing to revoke.See every permission
4. Where your data lives
Your account, learning, billing and support records, agent runs and reports are held in our database in the United States. Evidence files and cohort session recordings are stored with Amazon Web Services in the United States. Everything is encrypted in transit and at rest, and each workspace’s data can only be read inside that workspace.
5. Security controls
- Encryption. HTTPS only, with browsers told never to connect without it, and encryption at rest for the database and for evidence files.
- Sign-in. Password, Google or Microsoft sign-in, with optional two-step sign-in through an authenticator app. Our own staff cannot reach the admin console without one.
- Least access. Access within AURÄI is limited by role to the people who need it to run, support or secure the platform.
- Limits on repeated attempts. Sign-in, password reset and two-step codes are rate-limited.
- Audit trail. Account, role, evidence and billing actions, and every agent run, are recorded in an append-only audit log.
- Browser protections. The platform sends a content security policy and cannot be framed by other sites.
- Code changes. Every change reaches production through a pull request that passes automated checks, including a scan for leaked secrets, and the whole codebase is scanned again every week.
- Monitoring. Errors are reported with personal information removed before a report leaves the platform.
- Testing. In October 2026 we reviewed the platform and admin console and ran our own penetration tests against them, and fixed what we found.
6. Certifications and compliance
AURÄI does not currently hold SOC 2 or ISO/IEC 27001 certification, and the platform has not yet had an independent, third-party penetration test. The testing in section 5 was our own.
We handle personal information under the UK GDPR, the EU GDPR, the Nigeria Data Protection Act 2023, and the California and Texas privacy laws, as our privacy notices set out. Transfers out of the UK, the EEA and Nigeria rely on safeguards in our providers’ data processing terms, such as the EU Standard Contractual Clauses with the UK Addendum, or a provider’s certification under the EU-US Data Privacy Framework.
7. Subprocessors
These providers help us run the platform. Each processes data only for the purposes in our privacy notices and under data processing terms with us. The website’s own providers are listed in the Website Privacy Notice.
| Provider | What it does | Where |
|---|---|---|
| Vercel | Hosts the platform and runs its server code | United States |
| Cloudflare | DNS and network security in front of the platform | Global network |
| PlanetScale | Runs our database | United States |
| Amazon Web Services | Stores evidence files and cohort session recordings | United States |
| WorkOS | Sign-in, and verification and password-reset emails | United States |
| Stripe | Payments, tax, fraud screening, receipts and invoices | United States |
| OpenRouter and AI model providers | Run the AI models behind our agents, with zero data retention. Our Platform Privacy Notice names them. | United States or elsewhere |
| Bunny | Streams lesson videos | European Union, with a global delivery network |
| Resend | Sends course emails, two-step sign-in codes and security notices | United States |
| Upstash | Keeps the counters that limit repeated attempts | United States |
| Intercom | Runs the support chat | United States |
| PostHog | Product analytics and masked session recordings | European Union |
| Better Stack | Error reports | United States |
| Umami | Counts the steps to a purchase, on our own server | United States |
| Microsoft | Our email, and Microsoft Teams for live sessions | United States |
8. Report a security issue
If you think you have found a security weakness in AURÄI, or an incident affecting your data, email support@auraisystems.ai with what you found and how to reproduce it. Please do not access, change or delete anyone else’s data while you look, and give us a reasonable time to fix the issue before you tell anyone else.
If we learn of a breach affecting your information, we will tell you and the relevant regulator as the law requires.
9. Documents and contact
CyberOps AI Dynamics LLC (AURÄI)Houston, Texas, United States
support@auraisystems.ai