Caleta

AI Product Terms

Supplemental Terms for AI Features

Last Updated: March 17, 2026

1. Scope and Applicability

These AI Product Terms ("AI Terms") govern Your use of AI Features within the Caleta Service. They supplement and are incorporated into the Caleta Terms of Service ("TOS"). Capitalized terms not defined here have the meanings given in the TOS.

We may add or modify AI Features over time. Material changes to these AI Terms will be communicated at or before the time of change.

2. Definitions

"AI Features" means any product features or functionality within the Service that incorporate artificial intelligence or machine learning technology, including features powered by Third-Party AI Providers.

"Input" means any data, content, information, instructions, or prompts (in any format) submitted to, processed by, or made accessible to an AI Feature by or on behalf of You, Your authorized users, or Your end users.

"Output" means any result, response, recommendation, analysis, classification, or content (in any format) generated by an AI Feature based on Input.

"Third-Party AI Provider" means a third-party company whose AI models, APIs, or related technology Caleta uses to deliver AI Features.

"Training Data" means anonymized and de-identified data derived from Customer Data that is used to train, evaluate, or improve Caleta's proprietary AI models and algorithms.

Classification of Input and Output. Both Input and Output are classified as Customer Data under the TOS and are subject to all rights, restrictions, and obligations applicable to Customer Data, including confidentiality protections and data ownership.

3. Data Handling by Third-Party AI Providers

When You use AI Features, certain Input may be transmitted to Third-Party AI Providers for processing. Caleta maintains the following commitments regarding this data flow:

3.1 No Third-Party Model Training

Caleta does not permit any Third-Party AI Provider to use Your Customer Data (including Inputs and Outputs) to train, retrain, fine-tune, or otherwise improve that provider's own models or services. This prohibition is enforced through contractual obligations with each Third-Party AI Provider.

3.2 Data Retention at the AI Provider Layer

Caleta requires Third-Party AI Providers to implement zero or minimal data retention policies for Customer Data. Specifically:

  • Inputs and Outputs are processed ephemerally for inference and are not stored by Third-Party AI Providers beyond the time necessary to return a response.
  • Where a Third-Party AI Provider retains data for abuse monitoring or safety purposes, retention is limited to a maximum of 30 days and is subject to the same confidentiality commitments as all Customer Data.

3.3 Data Isolation

Customer Data processed through AI Features is logically isolated. Your data will not be commingled with another customer's data for any purpose other than generating aggregated, anonymized analytics that cannot be traced back to any individual customer.

4. Caleta's Use of Data for Model Improvement

This section describes how Caleta uses Customer Data to train and improve its own proprietary AI models and platform features. This is separate from and in addition to the restrictions on Third-Party AI Providers described in Section 3.

4.1 Training on Customer Data

Caleta uses anonymized and de-identified Customer Data to train and improve its proprietary AI models, algorithms, and platform features. This training enables Caleta to deliver increasingly accurate, relevant, and useful AI Features for all customers.

Before any Customer Data enters a training pipeline, Caleta applies a documented anonymization and sanitization process:

  • All directly identifying information (names, email addresses, account identifiers, and other personal identifiers) is stripped or replaced with non-reversible tokens.
  • Free-text fields are processed to identify and remove or mask personal information that is not relevant to model training.
  • The resulting Training Data cannot reasonably be traced back to any individual customer, user, or end user.
  • This anonymization pipeline runs entirely within Caleta's infrastructure. Training Data is not shared with Third-Party AI Providers.

4.2 Opt-Out Right

You may opt out of having Your Customer Data used for model training at any time by contacting support@caleta.us.

When You opt out:

  • New Customer Data generated after the opt-out will not be used for future model training.
  • Customer Data already incorporated into Training Data prior to the opt-out cannot be retroactively extracted from trained models. However, it will be excluded from all future training runs.
  • Opting out may result in loss of Your access to AI Features at the discretion of Caleta.

4.3 Enterprise Training Override

Customers on Enterprise-tier plans, or customers with individually negotiated agreements, may request a contractual commitment that defaults training to off, with opt-in only. This is available as an addendum to the Data Processing Agreement. Contact support@caleta.us for details.

4.4 Training Data Retention

Training Data (anonymized and de-identified) may be retained for up to five (5) years from the date of anonymization to support ongoing model development, evaluation, and safety improvement. After this period, Training Data will be deleted unless retention is required by law.

4.5 What Training Does Not Mean

To be clear, Caleta's use of data for training does not mean that:

  • Any individual customer's specific data, content, or business information will be surfaced to, visible to, or accessible by any other customer.
  • Customer Data is sold to, shared with, or made available to any third party for training or any other purpose.
  • Outputs generated for one customer will contain another customer's confidential information.
  • Caleta uses Customer Data for advertising, profiling, or any purpose other than improving the Service.

4.6 Aggregated Analytics

Separately, Caleta may use aggregated, de-identified Usage Data (as defined in the TOS) for product analytics, benchmarking, and reporting. Because Usage Data cannot reasonably be traced to any individual or customer, this use does not require opt-in or opt-out.

5. Transparency and Disclosure

5.1 Your Obligations to End Users

If You make AI Features available to Your own end users (for example, through a customer-facing application built on Caleta), You are responsible for ensuring that:

  • End users are informed that they are interacting with or receiving results from an AI system.
  • Outputs are clearly identifiable as AI-generated or AI-assisted.
  • Outputs are not falsely represented as human-generated.

6. Accuracy, Limitations, and Human Oversight

AI Features are provided on an "as-is" basis with respect to accuracy. Caleta does not warrant that Outputs will be accurate, complete, timely, or free from errors, hallucinations, bias, or offensive content.

You acknowledge and agree that:

  • Outputs should be reviewed by qualified personnel before any reliance, publication, or use in decision-making.
  • AI Features are not intended to substitute for professional judgment in any regulated domain, including legal, medical, financial, employment, or insurance contexts.
  • Different Inputs may produce different Outputs, and the same Input may produce different Outputs at different times.
  • Other customers may receive Outputs similar or identical to Yours based on their own Inputs. You have no exclusivity over Outputs.

7. AI-Specific Use Restrictions

In addition to the restrictions in the TOS and Acceptable Use Policy, You will not, and will not permit any third party to:

  • Use AI Features to develop AI or machine learning models, products, or services that compete with Caleta or any Third-Party AI Provider.
  • Attempt to extract, reverse-engineer, or infer the parameters, weights, training data, or architecture of any model underlying the AI Features.
  • Submit Inputs designed to manipulate, bypass, or test the safety or content-filtering mechanisms of the underlying models (including prompt injection, jailbreaking, or adversarial attacks).
  • Use Outputs in a manner that violates applicable law, the Acceptable Use Policy, or the terms of any Third-Party AI Provider.
  • Represent that AI-generated Outputs are human-generated in any context where such representation is misleading or prohibited by law.
  • Use AI Features to generate content that infringes third-party intellectual property rights, or to process data You are not authorized to submit.

8. Regulatory Compliance

You are responsible for evaluating whether Your use of AI Features complies with applicable laws and regulations, including data protection laws (GDPR, CCPA/CPRA), sector-specific regulations (HIPAA, FERPA, SOX), and AI-specific regulations (EU AI Act, California AI Transparency Act).

Caleta will provide commercially reasonable cooperation and documentation to assist You with compliance, including through the DPA, subprocessor disclosures, and the AI Trust page. However, compliance with applicable laws regarding Your use of Outputs and Your end users' data remains Your responsibility.

9. Changes to These AI Terms

We may update these AI Terms from time to time. Material changes will be communicated with you at or before the time of change. Your continued use of AI Features after the effective date of changes constitutes acceptance. If You do not agree, You may terminate Your Subscription in accordance with the TOS.

10. Contact

Questions about these AI Product Terms may be directed to: