AutoTrain ML Service Terms

Effective date: June 19, 2026

AutoTrain ML provides tools for dataset management, machine learning training, model evaluation, prediction testing, model export, dashboard history, and related automation workflows.

AutoTrain ML is not a data science consultant, legal advisor, compliance advisor, certification body, or decision maker. Outputs are intended to support your machine learning workflow and should be reviewed, validated, and tested by you before use in production or business decisions.

Accounts and Access

You may need an account to upload datasets, create training projects, run experiments, save model results, access testing tabs, manage exports, and view billing or usage history. You are responsible for keeping your credentials secure and for all activity under your account.

We may restrict, suspend, or remove access if we detect abuse, fraud, security risk, unusual system load, payment issues, or a violation of these terms.

Dataset Uploads

You are responsible for the datasets you upload and process through AutoTrain ML. You must have the right to upload, store, transform, train on, test, and export data submitted through the service.

• You should not upload unlawful, infringing, restricted, malicious, or third-party confidential content without permission.
• You should remove sensitive personal data unless you are legally allowed to process it and have appropriate safeguards.
• Dataset size, file type, row count, storage, and model limits may depend on your plan and system capacity.

Training, Testing, and Model Outputs

AutoTrain ML can run machine learning workflows such as preprocessing, supervised training, clustering, testing, prediction, visualization, and model export. You choose the datasets, target columns, algorithms, parameters, and testing inputs. The service may return metrics, charts, reports, prediction files, templates, trained model artifacts, and related metadata.

Machine learning outputs may contain mistakes, bias, overfitting, poor generalization, unexpected predictions, or unsuitable recommendations. AutoTrain ML does not guarantee accuracy, business performance, regulatory compliance, deployment readiness, or any particular result.

Processing and Infrastructure

AutoTrain ML may use the AutoTrain Engine, object storage, databases, queues, monitoring services, and secure routing to process training requests and serve the application. Jobs may be queued or delayed when workers are busy, files are large, dependencies are unavailable, or maintenance is required.

We may show maintenance notices, temporarily disable uploads or training, retry failed jobs, limit request rates, or pause access to protect reliability and data integrity.

Plans, Quotas, and Billing

AutoTrain ML has free and paid monthly subscription plans. Paid access begins after payment is confirmed and may renew monthly unless stated otherwise. The current self-service plans shown on the AutoTrain ML pricing page are:

MLXpert is a separate custom consulting option with tailored, scope-based pricing and is not part of the self-service monthly plan table above. Plan limits, worker availability, and subscription period can affect access to uploads, training, exports, model testing, API endpoints, and storage-related features. Unlimited testing means there is no per-test counter while the plan is active; fair-use and protective rate limits may still apply to keep AutoTrain Engine available for all users. If a paid subscription expires or an account moves to a lower plan, projects above the current limit are preserved but sealed until access is renewed or upgraded. Their inference API endpoints return an access error while unavailable.

Subscription dates, active plan status, invoices, and billing history may be shown in your account dashboard. Admins may manually adjust plan access when needed for support, testing, refunds, migration, or billing reconciliation.

Acceptable Use

You agree not to misuse AutoTrain ML, including by:

• attempting to bypass authentication, quotas, billing, or security controls;
• overloading infrastructure, creating abusive automated traffic, or disrupting other users;
• uploading malware, exploit payloads, or intentionally harmful files;
• violating intellectual property, privacy, export control, or other applicable laws;
• using outputs as the sole basis for high-risk legal, medical, financial, safety, or employment decisions.

Service Availability

AutoTrain ML is provided on an "as is" and "as available" basis to the maximum extent permitted by law. We aim to operate the service reliably, but we do not promise that every feature will be uninterrupted, error-free, compatible with every dataset, or available at all times.

Intellectual Property

You keep ownership of datasets and content you upload, subject to the rights you grant us to operate, store, process, transform, train, test, export, troubleshoot, and support your use of AutoTrain ML. AutoTrain ML, its interface, software, workflows, branding, and service design remain owned by their respective owners.

Refund Policy

For subscriptions, plan upgrades, and digital ML service access.

AutoTrain ML sells digital access to machine learning tools, training features, storage, testing, exports, and workflow capacity. Because access can be used immediately after purchase, refund eligibility depends on whether the plan has been used and whether training, exports, or other paid resources have already been consumed.

14-Day Refund Window

You may request a refund within 14 days of purchase for an accidental purchase or unused paid subscription, as long as paid features, training jobs, model exports, prediction outputs, or other paid resources have not been used or consumed.

Refunds We May Approve

• Duplicate charges for the same plan or package where only one purchase was intended.
• Accidental purchases reported within 14 days if paid features, training jobs, or exports have not been used.
• Technical failure where a paid feature was charged or consumed but no service was delivered, and we cannot reasonably restore, rerun, or credit the account.
• Unauthorized payment reports, subject to account review and payment provider verification.

Non-Refundable Situations

• Completed training jobs, generated model artifacts, prediction outputs, exports, or consumed paid capacity.
• Requests based only on model accuracy, dataset quality, chosen parameters, or expected business outcome.
• Abuse, fraud, policy violations, or account restriction caused by misuse.

Refunds may also be subject to payment provider terms, card network rules, and applicable consumer protection laws. To request a refund, contact us with your account email, order ID, provider reference, plan purchased, amount paid, payment date, and a short description of the issue. We aim to review refund requests within 5 business days.

Privacy Policy

How account, dataset, model, billing, and usage data are handled.

Information We May Collect

Depending on how you use AutoTrain ML, we may collect account information, uploaded datasets, file metadata, schema information, preprocessing choices, training parameters, model metrics, charts, prediction inputs, prediction outputs, exported artifacts, billing metadata, dashboard activity, technical logs, and analytics events.

How We Use Information

We use information to operate AutoTrain ML, authenticate users, store datasets, run training jobs, generate model artifacts, provide testing and exports, maintain account history, process payments, prevent abuse, troubleshoot errors, improve reliability, and communicate service notices.

Dataset and Model Processing

Uploaded datasets may be processed by AutoTrain ML infrastructure and machine learning workflows to parse files, infer schemas, preview rows, apply preprocessing, train models, calculate metrics, generate visualizations, export artifacts, and run predictions on new single or batch inputs.

Payment card details are handled by payment providers, not by AutoTrain ML. We store billing records needed to activate plans, reconcile purchases, prevent duplicate grants, handle refunds, and provide support.

Sharing and Providers

We do not sell your datasets, model artifacts, or account information. Standard technical data such as IP address, email address, payment metadata, browser logs, analytics events, and security events may pass through essential providers such as authentication, hosting, DNS, secure routing, payment processing, email, analytics, monitoring, or support tools.

Retention, Deletion, and Requests

We retain account, dataset, model, billing, and technical records for as long as needed to provide the service, maintain history, comply with legal obligations, resolve disputes, prevent abuse, and support troubleshooting. You may contact us to request access, correction, deletion, or export of information associated with your account.

Contact

Questions about checkout, refunds, account access, privacy, or service terms.

For support, billing, refund, account, or privacy questions, contact AutoTrain ML support. Policies may be updated from time to time, and the updated version will be posted on this page.