AI Disclaimer & Limitations Notice

Appraiser Guru™ by Tarnwick LLC · Last Updated: October 6, 2026

1. AI Technology Provider

The Appraiser Guru platform uses artificial intelligence technology provided by Anthropic, PBC ("AI Provider"). Specifically, the platform leverages Anthropic's Claude large language models (LLMs) to assist appraisers with analysis, content generation, and data processing tasks.

Claude is a conversational AI model trained on a broad range of text data and is designed to perform various language understanding and generation tasks.

2. AI Capabilities Within the Platform

2.1 Narrative Generation

The AI assists with generating draft narrative sections for appraisal reports, including: market conditions narratives, highest and best use analysis, site comments, condition and quality comments, and reconciliation narratives.

2.2 Guru Assistant

The AI answers your questions about the report you are working in and about the UAD specification, and helps route problems you report to our team.

2.3 Property Photo Reading

The AI reads a HUD data plate photo for a manufactured home's wind, thermal and roof-load zones, and estimates window area from front and rear elevation photos. It does not value, rate or judge the condition of a property from photographs.

2.4 Data Parsing and Structuring

The AI assists with: reading orders (pasted, or emailed to your order address), reading the first page of an appraisal PDF you upload to find the property, reading MLS listings and assessor or property-data pages you paste (populating structured fields), describing a market area from your MLS search map and criteria, and cleaning up the text of your own templates.

The AI does not suggest adjustment values and does not rank or score comparables. Adjustments, comparable selection and the opinion of value are yours.

3. Critical Limitations of AI Output

3.1 AI Does NOT Make Value Determinations

The Appraiser Guru AI does NOT determine property values, estimate a final appraised value, or make value conclusions. The AI assists with analysis and narrative generation, but the appraiser is solely responsible for reviewing all information, considering all relevant factors, and making the final value determination.

3.2 No Professional Judgment Substitution

AI-generated output does not substitute for the professional judgment, expertise, and legal responsibility of the licensed appraiser. The appraiser must independently evaluate all information, verify data accuracy, and ensure conclusions are sound.

3.3 No Guarantee of Accuracy

The Company and AI Provider make no warranty regarding the accuracy, completeness, or reliability of any AI-generated output. AI may produce factually incorrect statements (hallucinations), inappropriate adjustment values, logically inconsistent narratives, or conclusions that contradict evidence provided.

3.4 Not Professional Appraisal Advice

AI-generated output does not constitute professional appraisal advice, a professional appraisal opinion, or a substitute for appraiser expertise. Users must not rely on AI output without independent verification.

3.5 No Independent Data Access

The AI does not have independent access to live MLS data, public records databases, current market data beyond what users provide, or updated appraisal regulations. The AI can only analyze data that users submit.

4. Appraiser Responsibility and Duty

4.1 Independent Review Requirement

Every AI-generated output must be independently reviewed, verified, and edited by the licensed appraiser before inclusion in any appraisal report. The appraiser must verify that all statements of fact are accurate, all adjustments are appropriate and justified by market evidence, conclusions are logically consistent, and the narrative aligns with the appraiser's professional judgment.

Reliance on AI output without independent verification is at the appraiser's own risk and may violate USPAP standards. The appraiser is responsible for all content in the final report, regardless of source. AI errors do not excuse the appraiser from professional and regulatory obligations.

4.3 USPAP Compliance

Appraisers must comply with the Uniform Standards of Professional Appraisal Practice (USPAP) and all applicable state appraisal licensing regulations. Use of AI-generated content must be consistent with USPAP requirements regarding proper consideration of relevant data and disclosure of methodology.

4.4 Professional Liability

The appraiser is solely liable for all conclusions, statements, and opinions in the final appraisal report. The use of AI does not transfer professional responsibility to Tarnwick or the AI Provider. The appraiser cannot claim "the AI made this statement" as a defense for inaccurate content.

5. Data Handling and Privacy

5.1 Data Transmission to AI Provider

When an AI feature runs, the data that feature needs is transmitted to the AI Provider's API for processing; the AI Provider processes it on our behalf, as our service provider and subprocessor, only to provide the feature. Data a feature does not need is removed before it is sent. The features, what each one sends, and the four that run without a click (each of which you can switch off in Settings → Automatic AI) are listed in our Privacy Policy, Section 2.6.

5.2 What Anthropic's Published Terms Say

Our use of the AI Provider is governed by Anthropic's Commercial Terms of Service (effective June 17, 2025) and its Data Processing Addendum (effective February 24, 2025), as published by Anthropic. In those documents Anthropic states:

"Anthropic may not train models on Customer Content from Services." (Commercial Terms, Section B)
"Data submitted through the Services will be processed in accordance with the Anthropic Data Processing Addendum" (Commercial Terms, Section C)
"Customer is the controller and Anthropic is Customer’s processor." (Data Processing Addendum, Section B.1)
"Anthropic will only process Customer Personal Data to provide or maintain the Services, and in compliance with Customer’s documented instructions" (Data Processing Addendum, Section B.2)
Anthropic uses "industry standard encryption methods for protection of Customer Data, including a minimum of AES-256 for data at rest, and TLS1.2+ for data in transit." (Data Processing Addendum, Schedule 2, Section E)
"Anthropic will notify Customer in writing without undue delay, but in any event within 48 hours, after becoming aware of any Security Breach" (Data Processing Addendum, Section G.1)

Anthropic's Privacy Center article "How long do you store my organization's data?" (updated July 1, 2026) states: "For Anthropic API users, we automatically delete inputs and outputs on our backend within 30 days of receipt or generation, except" when a service with longer retention is used, when a different agreement such as zero data retention applies, when Anthropic needs to retain them longer to enforce its Usage Policy, or in compliance with the law. It also states that inputs and outputs flagged by Anthropic's automated trust and safety systems as violating its Usage Policy may be retained for up to 2 years, and that for certain models Anthropic designates it requires "limited data retention and review as a part of our safety work."

In these quotations "Customer" is Tarnwick and "Services" are Anthropic's services. They are Anthropic's own statements, which Anthropic may change; the linked pages are the current versions. If Anthropic materially changes its data handling, retention, or training policies in a way that affects how your data is processed, we will notify you within 30 days of becoming aware of the change.

5.3 Sensitive Information Caution

Users should not include sensitive information (SSNs, loan application details, confidential financial information of borrowers) in AI requests unless necessary for the appraisal function. We recommend redacting unnecessary sensitive data before submitting content to AI. Order reading is the one feature that necessarily sends a borrower's name and contact details and loan numbers, because the order form needs them; it removes quoted replies, earlier messages, signatures and confidentiality notices first, and its automatic form can be switched off in Settings → Automatic AI.

5.4 Data Processor List

A complete list of data processors, including Anthropic, is available in the Subprocessor List.

6. Known AI Limitations and Failure Modes

6.1 Hallucinations — Appraisal-Specific Examples

AI models can produce confident-sounding statements that are factually false or unsupported by provided data. In the appraisal context, hallucinations may include:

Fabricated comparable sales: The AI may reference a sale that never occurred, citing a plausible-sounding address, sale price, and date that cannot be verified in MLS or public records.
Incorrect zoning classifications: The AI may state a property is zoned R-1 Single Family when it is actually zoned R-2 or PUD, potentially affecting highest-and-best-use analysis.
Misidentified property features: The AI may describe a property as having a finished basement or attached garage when the photos or data show otherwise, leading to incorrect adjustments.
Inaccurate market trend statements: The AI may assert that the local market is "appreciating at 5% annually" when actual market data shows a different trend or declining conditions.
Wrong flood zone or regulatory information: The AI may state a property is in Zone X (minimal flood risk) when FEMA data shows it is in Zone AE (high risk), which could affect lender requirements and property value.

Always independently verify all factual claims in AI output against authoritative sources.

6.2 Other Known Limitations

Local Market Nuances: AI output reflects patterns in general training data and may not account for local market nuances, recent regulatory changes, or neighborhood-specific factors.
No Physical Inspection: The AI cannot perform physical inspections. Photo analysis is limited to what is visible in provided images.
Data Quality Dependency: AI output quality depends entirely on the quality of data provided. Inaccurate or incomplete data produces deficient output.
Bias and Fairness: AI models can reflect biases present in training data. Be vigilant about identifying and correcting biased conclusions, particularly regarding property characteristics, neighborhoods, or demographic factors.

7. Regulatory and Disclosure Considerations

Depending on applicable USPAP standards and state regulations, appraisers may be required to disclose that AI-assisted technology was used. Consult applicable USPAP Advisory Opinions and your state appraisal board for specific disclosure requirements.

Appraisers should clearly communicate to clients and lenders how AI was used in their work, what role it played, and that the appraiser independently verified all output.

8. Liability Disclaimer

8.1 Tarnwick's Disclaimer

Tarnwick LLC disclaims all warranties regarding AI output, including any implied warranties of merchantability, fitness for a particular purpose, or non-infringement. Tarnwick makes no warranty that AI output will be accurate, appropriate for regulatory standards, free from errors, or suitable for any particular use.

8.2 Limitation of Liability

Tarnwick shall not be liable for any damages arising from reliance on AI-generated content, including loss of income, professional liability or regulatory fines, third-party claims, or damaged professional reputation.

8.3 User Responsibility for Verification

Users acknowledge and accept full responsibility for independently verifying all AI-generated content before including it in any appraisal report or professional work product. The appraiser is the sole professional of record and bears full responsibility for the accuracy, completeness, and regulatory compliance of the final appraisal report, regardless of whether AI-assisted tools were used in its preparation.

9. Best Practices for AI Use

Always review AI output before relying on it.
Verify all factual statements independently.
Apply your professional judgment and market knowledge.
Edit AI-generated text to reflect your conclusions, not the AI's.
Be transparent with clients and lenders about your use of AI.
Ensure all content is consistent with USPAP and state licensing requirements.
Do not rely on AI for regulatory or legal compliance guidance.
Maintain professional liability insurance that covers technology-assisted work.

10. Contact Information

Tarnwick LLC
1625 Ridge Rd
Highland, MI 48356
Support: tarnwick.com/support
Email: info@tarnwick.com