AI Liability Waiver and Disclaimer

AI LIABILITY WAIVER AND DISCLAIMER — COUNSEL DRAFTING PACKET

This packet does not create an automatic waiver, warranty disclaimer, liability limit, indemnity, acceptance mechanism, or permission to deploy an AI system. It organizes the facts and decisions counsel needs to prepare product-specific terms for the actual system, users, jurisdictions, data, and uses.

1. Document Control

2. Release Gates

Do not publish or present user acceptance until each applicable gate is complete:

• ☐ The system profile and actual limitations in Sections 3 and 4 have been confirmed by the technical owner.
• ☐ Intended and prohibited uses are specific to this product rather than copied from a generic AI list.
• ☐ User types, locations, ages, languages, and transaction contexts are identified.
• ☐ Consumer, business, employee, contractor, student, patient, tenant, borrower, applicant, and public-sector uses have been classified separately.
• ☐ Consequential or safety-sensitive decisions and required human review are identified.
• ☐ Input, output, training, logging, retention, sharing, and deletion practices match the current product configuration.
• ☐ Confidential, personal, regulated, licensed, and third-party information has been mapped.
• ☐ Accessibility, language, notice, consent, and recordkeeping needs have been reviewed.
• ☐ Marketing, sales, documentation, demonstrations, and support statements have been checked against the proposed disclaimers.
• ☐ Warranty, remedy, liability, indemnity, governing-law, forum, arbitration, and fee choices have been approved for every covered user group and jurisdiction.
• ☐ The proposed acceptance method and evidence record have been reviewed.
• ☐ Insurance and upstream vendor terms have been compared with the risk allocated to users.
• ☐ Incident, complaint, appeal, correction, suspension, and change-notice procedures are operational.

3. System Profile

3.1 Product and Model Inventory

User-facing product description: [________________________________]

Inputs accepted: [________________________________]

Outputs produced: [________________________________]

External tools, retrieval sources, plugins, or data feeds: [________________________________]

Automated actions the system can take: [________________________________]

Actions requiring human approval: [________________________________]

3.2 Intended Uses

3.3 Excluded or Prohibited Uses

4. Verified Product Disclosures

Use only disclosures confirmed for the deployed system. Delete statements that do not match the product.

4.1 Performance and Output

Select and complete as applicable:

• ☐ Outputs may vary for the same or similar input. Product-specific explanation: [________________].
• ☐ Outputs may be inaccurate, incomplete, outdated, inconsistent, or unsupported. Known patterns: [________________].
• ☐ The system does not independently verify every output. Verification process offered to users: [________________].
• ☐ Performance differs by language, population, subject, format, or context. Known limits: [________________].
• ☐ The system uses external sources or tools whose availability and accuracy may vary. Sources or tools: [________________].
• ☐ The system may produce content similar to content available elsewhere. Review controls: [________________].
• ☐ Other verified limitation: [________________________________].

4.2 Currency and Updates

Knowledge or source currency: [________________________________]

Model, prompt, policy, or data update process: [________________________________]

User notice for material changes: [________________________________]

Version rollback or continuity process: [________________________________]

4.3 Human Review

Decisions for which output may be advisory only: [________________________________]

Required reviewer qualifications: [________________________________]

Escalation or second-review triggers: [________________________________]

User correction, appeal, or contest process: [________________________________]

5. User Groups, Jurisdictions, and Acceptance

5.1 Coverage Matrix

5.2 Acceptance Design

Counsel-selected presentation method:

• ☐ Signed agreement
• ☐ Click acceptance
• ☐ Account enrollment
• ☐ Order form or service agreement
• ☐ Workplace or institutional acknowledgment
• ☐ Informational notice only; no contractual acceptance asserted
• ☐ Other: [________________________________]

Text shown immediately before acceptance: [________________________________]

Required affirmative action: [________________________________]

Date, version, account, and evidence retained: [________________________________]

Method for withdrawal, rejection, or account closure: [________________________________]

Mere access, silence, or continued use is not treated by this drafting packet as acceptance. If counsel approves another mechanism, state the exact mechanism and supporting facts here: [________________________________].

6. User Responsibilities

Select only duties appropriate to the user, product, and transaction:

• ☐ Provide complete and accurate information required for the intended use.
• ☐ Review outputs before relying on, publishing, submitting, or acting on them.
• ☐ Use a qualified human reviewer for the uses identified in Section 4.3.
• ☐ Follow the product documentation, access controls, and security instructions.
• ☐ Do not submit information the user lacks authority to provide.
• ☐ Do not remove notices, safeguards, provenance information, or access restrictions.
• ☐ Do not use the system for the prohibited uses listed in Section 3.3.
• ☐ Report suspected harmful, unlawful, insecure, or materially incorrect behavior through: [________________].
• ☐ Preserve relevant inputs, outputs, settings, and versions when reporting a dispute or incident.
• ☐ Other negotiated responsibility: [________________________________].

These duties do not transfer the provider's own responsibilities to the user. The responsibility matrix in Schedule A must identify each party's operational role.

7. Data, Privacy, Confidentiality, and Security

7.1 Data-Flow Schedule

7.2 Required Product-Specific Statements

Whether inputs are used to train or improve any model: [________________________________]

Whether inputs or outputs are reviewed by people: [________________________________]

Whether content is shared with subprocessors or other providers: [________________________________]

Confidentiality treatment and exclusions: [________________________________]

Retention, deletion, export, and correction process: [________________________________]

Security measures represented to users: [________________________________]

Incident notice and support channel: [________________________________]

Do not say that information is confidential, anonymous, deleted, encrypted, not used for training, or kept in a particular location unless the current product and vendor configuration supports that statement.

8. Intellectual Property and Content

Counsel must complete the following for each input and output category:

This packet does not declare that an output is owned, copyrightable, original, noninfringing, exclusive, licensed, or available for a particular use. State the negotiated input license, output terms, attribution duties, provenance information, complaint process, and takedown or restriction process in Schedule B.

Open-source or source-license controls for generated code: [________________________________]

Brand, likeness, voice, publicity, or personality-right controls: [________________________________]

Complaint and challenged-content contact: [________________________________]

9. Professional, Consequential, and Safety-Sensitive Uses

Complete one row for each use that may affect health, safety, liberty, employment, housing, education, credit, insurance, access to services, legal rights, finances, or another material interest.

Select one for each proposed use:

• ☐ Prohibited.
• ☐ Permitted only as an internal support tool under the controls in Schedule C.
• ☐ Permitted after separate legal, safety, validation, and governance approval.
• ☐ Covered by a separate regulated-use agreement or policy identified here: [________________].

No “at your own risk” sentence in this packet substitutes for the product controls, review, notices, or approvals selected for a consequential use.

10. Warranty and Service Commitments

Reconcile this section with product documentation, sales statements, demonstrations, support terms, service levels, security commitments, and any order form.

Counsel-approved warranty text, if any, belongs in Schedule D. This packet supplies no default “as is,” merchantability, fitness, availability, accuracy, or noninfringement disclaimer.

11. Liability and Remedies Matrix

Do not apply one cap or exclusion to every claim, person, or loss. Complete the matrix after jurisdiction and transaction review.

Counsel-approved liability text belongs in Schedule E. If Schedule E is blank, this packet does not create a damages exclusion, dollar or fee cap, shortened claim period, exclusive remedy, risk assumption, release, or waiver.

12. Indemnity and Third-Party Claims

Any indemnity must identify the triggering claim, covered loss, indemnified persons, notice, control of defense, counsel, cooperation, settlement authority, conflicts, mitigation, exclusions, cap, and survival period.

Counsel-approved indemnity text belongs in Schedule F. No indemnity or defense duty applies if Schedule F is blank.

13. Suspension, Changes, and End of Service

Reasons for suspension or restriction: [________________________________]

Notice and opportunity to respond: [________________________________]

Emergency action and review process: [________________________________]

Export, transition, deletion, and account-closure steps: [________________________________]

Material-change notice period and delivery method: [________________________________]

Whether existing users must affirmatively accept a change: [________________________________]

This packet does not state that posting, silence, or continued use automatically accepts a modification. Counsel-approved change terms belong in Schedule G.

14. Complaints, Corrections, and Incidents

15. Governing Terms and Disputes

Selected governing law after counsel review: [________________________________]

Selected court and venue after counsel review: [________________________________]

Negotiation or escalation process: [________________________________]

Fees and costs: [________________________________]

This packet contains no arbitration agreement, jury-trial waiver, class or representative-action waiver, shortened limitation period, automatic prevailing-party fee, or exclusive forum. If approved, the complete negotiated term belongs in Schedule H with the covered users and disputes identified.

16. Final User-Facing Notice

After counsel completes the schedules, replace this section with a concise notice that a user can read before acceptance. The notice should identify, in plain language:

1. what the system does and does not do;
2. the verified limitations material to the intended use;
3. required human review and prohibited uses;
4. data, privacy, confidentiality, and security practices;
5. input and output terms;
6. support, correction, complaint, and appeal channels;
7. material warranty, remedy, liability, and indemnity terms approved by counsel;
8. suspension, change, and termination terms; and
9. governing terms and contact information.

Draft user-facing notice:

[________________________________]

17. Internal Approvals

Schedule A — Responsibility Matrix

[________________________________]

Schedule B — Input, Output, and Content Terms

[________________________________]

Schedule C — Consequential-Use Controls

[________________________________]

Schedule D — Counsel-Approved Warranty Terms

[________________________________]

Schedule E — Counsel-Approved Liability and Remedy Terms

[________________________________]

Schedule F — Counsel-Approved Indemnity Terms

[________________________________]

Schedule G — Suspension, Change, and End-of-Service Terms

[________________________________]

Schedule H — Dispute Terms

[________________________________]

Schedule I — Acceptance Evidence and Version Record