Privacy Policy

1. Who this covers

ExVista is a business service. Our customers are teams that scan, certify, and deploy machine-learning models. This policy covers the people who use the Services on behalf of those teams and visitors to our website. If you use the Services through an organization, that organization controls the account and the content in it, and its own policies may also apply.

2. What we collect

Account information

• Email address. Your sign-in name. Passwords are handled by Amazon Cognito; we never see or store the password itself.
• Google sign-in. If you choose “Continue with Google”, Google sends us your email address, whether Google has verified it, and a Google account identifier. We do not receive your Google password and we request no access to Gmail, Drive, Calendar, or any other Google data. If a password account already exists for the same verified email, the Google sign-in is joined to it so you keep one account.
• Two-factor authentication. If you enroll an authenticator app, Cognito stores the shared secret needed to check your codes.
• Access level. Whether your account is in a partner or administrator group, which controls which features you can use.

Content you submit for scanning

The Services exist to examine models, so they necessarily process the models you point them at (“Customer Content”):

• Model weights and files you upload directly, or that we download from a Hugging Face repository you name, or that we read from an Amazon S3 bucket you connect. A connected bucket is read through a role you create in your own AWS account, scoped to ExVista by an external ID. It is read-only and we hold no standing credentials for it.
• Scan results. The report for each scan: the verdict, confidence, candidate trigger phrases and target behaviors the detector reconstructed, excerpts of text the model produced while being probed, coverage statistics, and fingerprints of the model (hashes of its weight bytes, of its outputs on a fixed probe set, and of pooled internal activations).
• Operational metadata about each scan: the model name and size you entered, which GPU class ran it, how long it took, its stage-by-stage progress, and any error message.
• Temporary checkpoints. While a scan runs, intermediate state, including raw model outputs, is checkpointed so an interrupted scan can resume. Checkpoints are deleted when the scan completes and expire automatically within three days regardless.

Projects, devices, and deployments

• Names and labels you give workspaces, projects, and devices, and the platform string a device reports (for example the board and operating system).
• Device keys. Each enrolled device gets a secret key shown once. We store only a hash of it and its last four characters.
• Deployment telemetry. When a device checks in, we record the time, which deployment it pulled or loaded, the fingerprint of the model bytes it is running, and a short status string. These form an append-only provenance log of which model ran on which device, when.

Command-line tool

Signing in from the terminal uses a one-time code you approve in the browser. We record which account approved a code and when. The CLI keeps its sign-in tokens in a file in your home directory on your own machine; we do not collect the contents of your terminal or your filesystem.

Website analytics

Our website and console use Amazon CloudWatch RUM to understand how the product is used and to catch errors. It records page views, page-load timing, JavaScript errors, and clicks on links and buttons (the label of the element and the page it was on, never the contents of form fields). It sets two cookies that identify your browser session and your browser across visits, used only for these analytics. It also captures technical details such as browser and operating system type.

Communications

The demo request form opens your own email client; we receive whatever you choose to send. If you write to us, we keep the correspondence so we can respond and follow up.

Server logs

Like any hosted service, our infrastructure logs requests, including IP addresses, timestamps, and request details, for security, debugging, and capacity planning.

3. How we use it

• To provide the Services: run scans, produce reports, fingerprint and re-verify models, decide whether a deployment may be served to a device, and let you sign in from the browser and terminal.
• To keep the Services secure: detect abuse, enforce per-account isolation, and investigate incidents.
• To support you and respond to your requests.
• To improve the product. We use Customer Content only to provide the Services to you, to operate and secure them, and, in aggregated or de-identified form that does not identify you or your models, to improve detection quality. We do not use your model weights to train models of our own.
• To meet legal obligations and enforce our Terms of Service.

We do not sell personal information, and we do not use it for third-party advertising.

Our use of information received from Google APIs adheres to the Google API Services User Data Policy, including its Limited Use requirements.

4. Who we share it with

• Amazon Web Services hosts everything: the website, the console, storage for uploads and reports, the GPU workers that run scans, authentication (Cognito), and analytics (CloudWatch RUM). Data is stored in the US East (Ohio) region.
• Google, only if you sign in with Google. Google’s own privacy policy governs what Google does on its side.
• Hugging Face, only when you ask us to scan a model hosted there. Our workers fetch the repository you named; Hugging Face sees that request, not your ExVista account.
• Your own AWS account, when you connect a bucket. The role and its permissions are yours to change or revoke at any time.
• Legal and safety. We may disclose information if required by law or if we believe it is necessary to protect the rights, safety, or property of ExVista, our customers, or others.
• Business transfers. If ExVista is acquired or merges, this data may transfer to the successor, who will remain bound by this policy.

We use no advertising networks and no third-party trackers.

5. How long we keep it

• Account information for as long as your account exists.
• Uploads, scan reports, and fingerprints until you delete the scan or your account. You can delete individual scans from the console.
• Scan checkpoints are deleted when a scan finishes and expire automatically within three days in every case.
• Device telemetry and the provenance log for as long as the project exists, because their purpose is an audit trail.
• Analytics events are kept in raw form for 30 days and afterwards only as aggregate counts.
• Server logs for a limited operational period.

6. How we protect it

Data is encrypted in transit and at rest. Each account’s uploads, reports, scans, devices, and connected buckets are isolated to that account; a deployment can only reuse a scan the same account ran. Device keys are stored only as hashes. Connected buckets are read through customer-owned, read-only roles. Terminal sign-in never types a password into the terminal. You can add two-factor authentication to a password account from the console’s security page. No system is perfectly secure, and we cannot guarantee that unauthorized access will never occur.

7. Your choices and rights

• Review and update your sign-in email and password from the security page of the console.
• Delete scans, projects, and devices from the console.
• Delete your account and its content by emailing us at babs@exvistatechnologies.com. We will confirm the request and remove your data, except what we must keep for legal or security reasons.
• Analytics cookies can be blocked or cleared in your browser; the Services work without them.
• Google sign-in can be disconnected from your Google account’s third-party access settings at any time.

Depending on where you live, you may have rights to access, correct, export, or delete your personal information, or to object to or restrict certain processing. Email us and we will honor them as the law requires. We will not treat you differently for exercising them.

8. Where data lives

The Services are operated from the United States and your information is stored there. If you use them from elsewhere, you are sending your information to the United States, where privacy law may differ from your own jurisdiction’s.

9. Children

The Services are for business use by adults. We do not knowingly collect information from anyone under 18.

10. Changes to this policy

When we change this policy we will update the effective date above. For material changes we will also notify account holders by email before the change takes effect.

11. Contact

Questions, requests, or concerns about privacy: babs@exvistatechnologies.com.