AI Services Addendum

This AI Services Addendum ("AI Addendum") forms part of the Chat4U Agreement between Pirsum-Mi Ltd, operating as Chat4U, and Customer. It applies to Customer's use of AI-enabled features in the Service, including retrieval-based responses, prompt orchestration, model generation, embeddings, reranking, classification, summarization, and Pro Agent features.

Capitalized terms not defined here have the meanings given in the Terms of Service, the AUP, the DPA, or the applicable Order.

1. Scope

This AI Addendum applies to:

• standard AI chat features;
• retrieval-augmented generation and related retrieval features;
• prompt-driven classification or extraction features;
• model-generated summaries or responses;
• embeddings and related semantic processing;
• optional moderation, safety, or quality layers;
• Pro Agent functionality; and
• related AI provider integrations used by Chat4U to operate the Service.

2. Definitions

For purposes of this AI Addendum:

"AI Feature" means a Service feature that uses one or more machine-learning or model-based systems to generate, transform, classify, rank, summarize, retrieve, or act on content.

"Customer Data" has the meaning given in the Terms of Service and DPA.

"Input" means prompts, instructions, messages, configurations, data sources, retrieved content, files, text, tool inputs, and other information submitted to or used by an AI Feature.

"Output" means text, classifications, summaries, structured fields, actions, suggestions, or other content generated or returned by an AI Feature.

"Provider" means a third-party model, embedding, reranking, hosting, safety, or related provider used by Chat4U or configured by Customer.

3. Customer Responsibility for Inputs and Use

Customer is responsible for:

• the legality, appropriateness, and authorization of Inputs;
• how Inputs are configured;
• which data sources are used;
• whether outputs are reviewed before use;
• the context in which Output is shown or acted upon; and
• whether Customer's deployment is appropriate for its users and legal obligations.

Chat4U operates the AI pipeline for the Service but does not assume responsibility for Customer's business, compliance, or decision-making obligations.

4. AI Limitations

AI Features are probabilistic and may produce Outputs that are inaccurate, incomplete, outdated, misleading, biased, non-unique, or inappropriate.

Even when an Output is grounded in retrieved content, the Service may:

• retrieve the wrong source;
• fail to retrieve the best source;
• retrieve stale or incomplete source material;
• misread context;
• summarize imperfectly;
• classify incorrectly; or
• generate an answer that is still wrong.

Customer must not treat Output as guaranteed fact.

5. Optional Guardrails and Quality Features

Some safety, moderation, or quality controls may be configurable, optional, dependent on plan, dependent on provider, or disabled.

Guardrails may be enabled or disabled at the platform or agent level. Hallucination checks, if offered, are optional features and may be disabled by default. A configured or available safety feature is not a warranty of correctness, legality, or fitness for use.

Customer remains responsible for determining whether review, escalation, or additional controls are needed.

6. Human Review and Escalation

Customer must implement human review appropriate to its use case.

At a minimum, Customer must not rely on Output without appropriate human review where the Output could affect:

• health;
• legal rights or obligations;
• finances;
• safety;
• access to services;
• employment;
• housing;
• education;
• insurance;
• credit;
• benefits; or
• similar material interests.

Customer is responsible for setting escalation paths, fallback behavior, and review procedures that fit its use case.

7. Output Rights and Restrictions

Subject to this Agreement, applicable law, Provider terms, and any rights in underlying materials, Customer may use Output generated for Customer.

Output may not be unique. Similar or identical output may be generated for Chat4U, other customers, or third parties.

Chat4U does not guarantee ownership, exclusivity, non-infringement, merchantability, or fitness of Output.

To the extent, if any, that Chat4U acquires rights in Output that do not automatically vest in Customer, Chat4U assigns those rights to Customer to the extent legally possible, subject to:

• Customer's compliance with the Agreement;
• rights in underlying Inputs and third-party materials;
• Provider terms;
• Chat4U's rights in the Service and Company Usage Data; and
• applicable law.

Nothing in this AI Addendum transfers to Customer any right in:

• the Service itself;
• underlying models;
• system prompts or platform prompts used by Chat4U generally;
• provider technology;
• embeddings infrastructure;
• retrieval logic;
• logs or diagnostics maintained by Chat4U as controller; or
• Company Usage Data.

8. No Cross-Customer Model Training by Chat4U

Chat4U does not use Customer Inputs or Customer Outputs to train a cross-customer or general-purpose model owned by Chat4U.

This statement applies only to Chat4U's own training practices. It does not itself guarantee how third-party Providers process data. Provider processing is governed by:

• the Agreement;
• the DPA where applicable;
• Chat4U's then-current subprocessor disclosures; and
• relevant Provider terms.

Customer acknowledges that Provider practices may differ by provider and by feature.

9. Provider Changes

Chat4U may add, remove, replace, or reconfigure Providers used to operate AI Features.

Reasons may include:

• availability;
• product performance;
• safety;
• pricing;
• technical compatibility;
• abuse prevention;
• supportability; or
• legal or compliance considerations.

Customer-configured provider keys or integrations remain Customer's responsibility.

10. AI Disclosure Duties

Customer is responsible for making any end-user disclosures required by law or reasonably required by the context of use, including where a visitor is interacting with an AI system rather than a human.

Customer must not configure the Service to deceptively misrepresent AI outputs as human-authored where such a representation would be unlawful or misleading.

11. High-Risk Use Prohibition

For purposes of this AI Addendum, "High-Risk Use" means providing professional medical, legal, or financial advice, or making or materially informing decisions about employment, housing, credit, insurance, education access, benefits eligibility, essential services, biometric identification or verification, safety-critical matters, government functions, defense activities, or similar high-impact outcomes.

High-Risk Use is prohibited unless a separately signed Enterprise agreement expressly authorizes the exact use.

Ordinary FAQ, support, and navigation assistance are permitted only when they do not provide such advice and do not make or materially inform such decisions.

Customer must not assume that:

• a general contract;
• a negotiated commercial relationship;
• a support conversation;
• customer-side oversight;
• internal policy;
• optional guardrails;
• output review; or
• a request for exception review

authorizes High-Risk Use.

Any possible exception is not self-service, is not guaranteed to be available, and is not created by requesting review. Customer may contact support@chat4u.ai to ask whether an Enterprise exception program is available for a specific proposed use, but the request itself does not authorize use and does not guarantee review or approval.

12. No Professional Advice

The Service is not a medical, legal, or financial professional.

Customer must not present AI Features or Output as:

• medical advice;
• legal advice;
• financial advice;
• a substitute for qualified professional judgment; or
• an official decision-maker.

13. Pro Agent Terms

13.1 Operational Risk

Pro Agent features can request browser actions and inspect browser content. They may create heightened operational, legal, and privacy risk.

Customer is responsible for whether and how Pro Agent is enabled and supervised.

13.2 Absolute Secrets Prohibition

Customer must not use Pro Agent to collect, request, store, process, reveal, submit, or exfiltrate:

• passwords;
• passcodes;
• one-time authentication codes;
• multi-factor authentication secrets;
• payment-card numbers;
• card verification values;
• bank credentials; or
• similarly sensitive secrets.

This rule is absolute. No exception is created by consent, internal workflow, customer policy, or requested use.

13.3 Safeguards Do Not Remove Responsibility

Chat4U may provide tool approval flows, deny selectors, limited redaction, or other protective measures. These measures reduce risk but do not guarantee that every harmful or unauthorized action will be prevented.

14. Sensitive Data

Customer must not use AI Features for PHI or other regulated or highly sensitive data unless Chat4U actually offers and signs a separate written agreement covering the exact use and data type.

This AI Addendum does not state that Chat4U currently offers a BAA.

15. Relationship to Other Documents

This AI Addendum supplements the Terms of Service, the AUP, and the DPA.

If there is a conflict:

• the Terms of Service govern general commercial terms;
• the AUP governs prohibited conduct and prohibited use cases;
• the DPA governs Customer Data processing terms where Chat4U acts as processor or sub-processor; and
• this AI Addendum governs AI-specific risk allocation and usage terms.

Nothing in this AI Addendum narrows the AUP's prohibited-use restrictions.